1 systemd System and Service Manager
5 * A new tool systemd-sysusers has been added. This tool
6 creates system users and groups in /etc/passwd and
7 /etc/group, based on static declarative system user/group
8 definitions in /usr/lib/sysusers.d/. This is useful to
9 enable factory resets and volatile systems that boot up with
10 an empty /etc directory, and thus need system users and
11 groups created during early boot. systemd now also ships
12 with two default sysusers.d/ files for the most basic
13 users and groups systemd and the core operating system
16 * A new tmpfiles snippet has been added that rebuilds the
17 essential files in /etc on boot, should they be missing.
19 * A directive for ensuring automatic clean-up of
20 /var/cache/man/ has been removed from the default
21 configuration. This line should now be shipped by the man
22 implementation. The necessary change has been made to the
23 man-db implementation. Note that you need to update your man
24 implementation to one that ships this line, otherwise no
25 automatic clean-up of /var/cache/man will take place.
27 * A new condition ConditionNeedsUpdate= has been added that
28 may conditionalize services to only run when /etc or /var
29 are "older" than the vendor operating system resources in
30 /usr. This is useful for reconstructing or updating /etc
31 after an offline update of /usr or a factory reset, on the
32 next reboot. Services that want to run once after such an
33 update or reset should use this condition and order
34 themselves before the new systemd-update-done.service, which
35 will mark the two directories as fully updated. A number of
36 service files have been added making use of this, to rebuild
37 the udev hardware database, the journald message catalog and
38 dynamic loader cache (ldconfig). The systemd-sysusers tool
39 described above also makes use of this now. With this in
40 place it is now possible to start up a minimal operating
41 system with /etc empty cleanly. For more information on the
42 concepts involved see this recent blog story:
44 http://0pointer.de/blog/projects/stateless.html
46 * A new system group "input" has been introduced, and all
47 input device nodes get this group assigned. This is useful
48 for system-level software to get access to input devices. It
49 complements what is already done for "audio" and "video".
51 * systemd-networkd learnt minimal DHCPv4 server support in
52 addition to the existing DHCPv4 client support. It also
53 learnt DHCPv6 client and IPv6 Router Solicitation client
54 support. The DHCPv4 client gained support for static routes
55 passed in from the server. Note that the [DHCPv4] section
56 known in older systemd-networkd versions has been renamed to
57 [DHCP] and is now also used by the DHCPv6 client. Existing
58 .network files using settings of this section should be
59 updated, though compatibility is maintained. Optionally, the
60 client hostname may now be sent to the DHCP server.
62 * networkd gained support for vxlan virtual networks as well
63 as tun/tap and dummy devices.
65 * networkd gained support for automatic allocation of address
66 ranges for interfaces from a system-wide pool of
67 addresses. This is useful for dynamically managing a large
68 number of interfaces with a single network configuration
69 file. In particular this is useful to easily assign
70 appropriate IP addresses to the veth links of a large number
73 * RPM macros for processing sysusers, sysctl and binfmt
74 drop-in snippets at package installation time have been
77 * The /etc/os-release file should now be placed in
78 /usr/lib/os-release. The old location is automatically
79 created as symlink. /usr/lib is the more appropriate
80 location of this file, since it shall actually describe the
81 vendor operating system shipped in /usr, and not the
82 configuration stored in /etc.
84 * .mount units gained a new boolean SloppyOptions= setting
85 that maps to mount(8)'s -s option which enables permissive
86 parsing of unknown mount options.
88 * tmpfiles learnt a new "L+" directive which creates a symlink
89 but (unlike "L") deletes a pre-existing file first, should
90 it already exist and not already be the correct
91 symlink. Similar, "b+", "c+" and "p+" directives have been
92 added as well, which create block and character devices, as
93 well as fifos in the filesystem, possibly removing any
94 pre-existing files of different types.
96 * For tmpfiles' "L", "L+", "C" and "C+" directives the final
97 'argument' field (which so far specified the source to
98 symlink/copy the files from) is now optional. If omitted the
99 same file os copied from /usr/share/factory/ suffixed by the
100 full destination path. This is useful for populating /etc
101 with essential files, by copying them from vendor defaults
102 shipped in /usr/share/factory/etc.
104 * A new command "systemctl preset-all" has been added that
105 applies the service preset settings to all installed unit
106 files. A new switch --preset-mode= has been added that
107 controls whether only enable or only disable operations
110 * A new command "systemctl is-system-running" has been added
111 that allows checking the overall state of the system, for
112 example whether it is fully up and running.
114 * When the system boots up with an empty /etc, the equivalent
115 to "systemctl preset-all" is executed during early boot, to
116 make sure all default services are enabled after a factory
119 * systemd now contains a minimal preset file that enables the
120 most basic services systemd ships by default.
122 * Unit files' [Install] section gained a new DefaultInstance=
123 field for defining the default instance to create if a
124 template unit is enabled with no instance specified.
126 * A new passive target cryptsetup-pre.target has been added
127 that may be used by services that need to make they run and
128 finish before the first LUKS cryptographic device is set up.
130 * The /dev/loop-control and /dev/btrfs-control device nodes
131 are now owned by the "disk" group by default, opening up
132 access to this group.
134 * systemd-coredump will now automatically generate a
135 stack trace of all core dumps taking place on the system,
136 based on elfutils' libdw library. This stack trace is logged
139 * systemd-coredump may now optionally store coredumps directly
140 on disk (in /var/lib/systemd/coredump, possibly compressed),
141 instead of storing them unconditionally in the journal. This
142 mode is the new default. A new configuration file
143 /etc/systemd/coredump.conf has been added to configure this
144 and other parameters of systemd-coredump.
146 * coredumpctl gained a new "info" verb to show details about a
147 specific coredump. A new switch "-1" has also been added
148 that makes sure to only show information about the most
149 recent entry instead of all entries. Also, as the tool is
150 generally useful now the "systemd-" prefix of the binary
151 name has been removed. Distributions that want to maintain
152 compatibility with the old name should add a symlink from
153 the old name to the new name.
155 * journald's SplitMode= now defaults to "uid". This makes sure
156 that unprivileged users can access their own coredumps with
157 coredumpctl without restrictions.
159 * New kernel command line options "systemd.wants=" (for
160 pulling an additional unit during boot), "systemd.mask="
161 (for masking a specific unit for the boot), and
162 "systemd.debug-shell" (for enabling the debug shell on tty9)
163 have been added. This is implemented in the new generator
164 "systemd-debug-generator".
166 * systemd-nspawn will now by default filter a couple of
167 syscalls for containers, among them those required for
168 kernel module loading, direct x86 IO port access, swap
169 management, and kexec. Most importantly though
170 open_by_handle_at() is now prohibited for containers,
171 closing a hole similar to a recently discussed vulnerability
172 in docker regarding access to files on file hierarchies the
173 container should normally not have access to. Note that for
174 nspawn we generally make no security claims anyway (and
175 this is explicitly documented in the man page), so this is
176 just a fix for one of the most obvious problems.
178 * A new man page file-hierarchy(7) has been added that
179 contains a minimized, modernized version of the file system
180 layout systemd expects, similar in style to the FHS
181 specification or hier(5). A new tool systemd-path(1) has
182 been added to query many of these paths for the local
185 * Automatic time-based clean-up of $XDG_RUNTIME_DIR is no
186 longer done. Since the directory now has a per-user size
187 limit, and is cleaned on logout this appears unnecessary,
188 in particular since this now brings the lifecycle of this
189 directory closer in line with how IPC objects are handled.
191 * systemd.pc now exports a number of additional directories,
192 including $libdir (which is useful to identify the library
193 path for the primary architecture of the system), and a
194 couple of drop-in directories.
196 * udev's predictable network interface names now use the dev_port
197 sysfs attribute, introduced in linux 3.15 instead of dev_id to
198 distinguish between ports of the same PCI function. dev_id should
199 only be used for ports using the same HW address, hence the need
202 * machined has been updated to export the OS version of a
203 container (read from /etc/os-release and
204 /usr/lib/os-release) on the bus. This is now shown in
205 "machinectl status" for a machine.
207 * A new service setting RestartForceExitStatus= has been
208 added. If configured to a set of exit signals or process
209 return values, the service will be restarted when the main
210 daemon process exits with any of them, regardless of the
213 * systemctl's -H switch for connecting to remote systemd
214 machines has been extended so that it may be used to
215 directly connect to a specific container on the
216 host. "systemctl -H root@foobar:waldi" will now connect as
217 user "root" to host "foobar", and then proceed directly to
218 the container named "waldi". Note that currently you have to
219 authenticate as user "root" for this to work, as entering
220 containers is a privileged operation.
222 Contributions from: Andreas Henriksson, Benjamin Steinwender,
223 Carl Schaefer, Christian Hesse, Colin Ian King, Cristian
224 Rodríguez, Daniel Mack, Dave Reisner, David Herrmann, Eugene
225 Yakubovich, Filipe Brandenburger, Frederic Crozat, Hristo
226 Venev, Jan Engelhardt, Jonathan Boulle, Kay Sievers, Lennart
227 Poettering, Luke Shumaker, Mantas Mikulėnas, Marc-Antoine
228 Perennou, Marcel Holtmann, Michael Marineau, Michael Olbrich,
229 Michał Bartoszkiewicz, Michal Sekletar, Patrik Flykt, Ronan Le
230 Martret, Ronny Chevalier, Ruediger Oertel, Steven Noonan,
231 Susant Sahani, Thadeu Lima de Souza Cascardo, Thomas Hindoe
232 Paaboel Andersen, Tom Gundersen, Tom Hirst, Umut Tezduyar
233 Lindskog, Uoti Urpala, Zbigniew Jędrzejewski-Szmek
235 -- Berlin, 2014-07-03
239 * As an experimental feature, udev now tries to lock the
240 disk device node (flock(LOCK_SH|LOCK_NB)) while it
241 executes events for the disk or any of its partitions.
242 Applications like partitioning programs can lock the
243 disk device node (flock(LOCK_EX)) and claim temporary
244 device ownership that way; udev will entirely skip all event
245 handling for this disk and its partitions. If the disk
246 was opened for writing, the close will trigger a partition
247 table rescan in udev's "watch" facility, and if needed
248 synthesize "change" events for the disk and all its partitions.
249 This is now unconditionally enabled, and if it turns out to
250 cause major problems, we might turn it on only for specific
251 devices, or might need to disable it entirely. Device Mapper
252 devices are excluded from this logic.
254 * We temporarily dropped the "-l" switch for fsck invocations,
255 since they collide with the flock() logic above. util-linux
256 upstream has been changed already to avoid this conflict,
257 and we will readd "-l" as soon as util-linux with this
258 change has been released.
260 * The dependency on libattr has been removed. Since a long
261 time, the extended attribute calls have moved to glibc, and
262 libattr is thus unnecessary.
264 * Virtualization detection works without priviliges now. This
265 means the systemd-detect-virt binary no longer requires
266 CAP_SYS_PTRACE file capabilities, and our daemons can run
267 with fewer privileges.
269 * systemd-networkd now runs under its own "systemd-network"
270 user. It retains the CAP_NET_ADMIN, CAP_NET_BIND_SERVICE,
271 CAP_NET_BROADCAST, CAP_NET_RAW capabilities though, but
272 loses the ability to write to files owned by root this way.
274 * Similar, systemd-resolved now runs under its own
275 "systemd-resolve" user with no capabilities remaining.
277 * Similar, systemd-bus-proxyd now runs under its own
278 "systemd-bus-proxy" user with only CAP_IPC_OWNER remaining.
280 * systemd-networkd gained support for setting up "veth"
281 virtual ethernet devices for container connectivity, as well
282 as GRE and VTI tunnels.
284 * systemd-networkd will no longer automatically attempt to
285 manually load kernel modules necessary for certain tunnel
286 transports. Instead, it is assumed the kernel loads them
287 automatically when required. This only works correctly on
288 very new kernels. On older kernels, please consider adding
289 the kernel modules to /etc/modules-load.d/ as a work-around.
291 * The resolv.conf file systemd-resolved generates has been
292 moved to /run/systemd/resolve/. If you have a symlink from
293 /etc/resolv.conf, it might be necessary to correct it.
295 * Two new service settings, ProtectedHome= and ProtectedSystem=,
296 have been added. When enabled, they will make the user data
297 (such as /home) inaccessible or read-only and the system
298 (such as /usr) read-only, for specific services. This allows
299 very light-weight per-service sandboxing to avoid
300 modifications of user data or system files from
301 services. These two new switches have been enabled for all
302 of systemd's long-running services, where appropriate.
304 * Socket units gained new SocketUser= and SocketGroup=
305 settings to set the owner user and group of AF_UNIX sockets
306 and FIFOs in the file system.
308 * Socket units gained a new RemoveOnStop= setting. If enabled,
309 all FIFOS and sockets in the file system will be removed
310 when the specific socket unit is stopped.
312 * Socket units gained a new Symlinks= setting. It takes a list
313 of symlinks to create to file system sockets or FIFOs
314 created by the specific Unix sockets. This is useful to
315 manage symlinks to socket nodes with the same life-cycle as
318 * The /dev/log socket and /dev/initctl FIFO have been moved to
319 /run, and have been replaced by symlinks. This allows
320 connecting to these facilities even if PrivateDevices=yes is
321 used for a service (which makes /dev/log itself unavailable,
322 but /run is left). This also has the benefit of ensuring
323 that /dev only contains device nodes, directories and
324 symlinks, and nothing else.
326 * sd-daemon gained two new calls sd_pid_notify() and
327 sd_pid_notifyf(). They are similar to sd_notify() and
328 sd_notifyf(), but allow overriding of the source PID of
329 notification messages if permissions permit this. This is
330 useful to send notify messages on behalf of a different
331 process (for example, the parent process). The
332 systemd-notify tool has been updated to make use of this
333 when sending messages (so that notification messages now
334 originate from the shell script invoking systemd-notify and
335 not the systemd-notify process itself. This should minimize
336 a race where systemd fails to associate notification
337 messages to services when the originating process already
340 * A new "on-abnormal" setting for Restart= has been added. If
341 set, it will result in automatic restarts on all "abnormal"
342 reasons for a process to exit, which includes unclean
343 signals, core dumps, timeouts and watchdog timeouts, but
344 does not include clean and unclean exit codes or clean
345 signals. Restart=on-abnormal is an alternative for
346 Restart=on-failure for services that shall be able to
347 terminate and avoid restarts on certain errors, by
348 indicating so with an unclean exit code. Restart=on-failure
349 or Restart=on-abnormal is now the recommended setting for
350 all long-running services.
352 * If the InaccessibleDirectories= service setting points to a
353 mount point (or if there are any submounts contained within
354 it), it is now attempted to completely unmount it, to make
355 the file systems truly unavailable for the respective
358 * The ReadOnlyDirectories= service setting and
359 systemd-nspawn's --read-only parameter are now recursively
360 applied to all submounts, too.
362 * Mount units may now be created transiently via the bus APIs.
364 * The support for SysV and LSB init scripts has been removed
365 from the systemd daemon itself. Instead, it is now
366 implemented as a generator that creates native systemd units
367 from these scripts when needed. This enables us to remove a
368 substantial amount of legacy code from PID 1, following the
369 fact that many distributions only ship a very small number
370 of LSB/SysV init scripts nowadays.
372 * Priviliged Xen (dom0) domains are not considered
373 virtualization anymore by the virtualization detection
374 logic. After all, they generally have unrestricted access to
375 the hardware and usually are used to manage the unprivileged
378 * systemd-tmpfiles gained a new "C" line type, for copying
379 files or entire directories.
381 * systemd-tmpfiles "m" lines are now fully equivalent to "z"
382 lines. So far, they have been non-globbing versions of the
383 latter, and have thus been redundant. In future, it is
384 recommended to only use "z". "m" has hence been removed
385 from the documentation, even though it stays supported.
387 * A tmpfiles snippet to recreate the most basic structure in
388 /var has been added. This is enough to create the /var/run →
389 /run symlink and create a couple of structural
390 directories. This allows systems to boot up with an empty or
391 volatile /var. Of course, while with this change, the core OS
392 now is capable with dealing with a volatile /var, not all
393 user services are ready for it. However, we hope that sooner
394 or later, many service daemons will be changed upstream so
395 that they are able to automatically create their necessary
396 directories in /var at boot, should they be missing. This is
397 the first step to allow state-less systems that only require
398 the vendor image for /usr to boot.
400 * systemd-nspawn has gained a new --tmpfs= switch to mount an
401 empty tmpfs instance to a specific directory. This is
402 particularly useful for making use of the automatic
403 reconstruction of /var (see above), by passing --tmpfs=/var.
405 * Access modes specified in tmpfiles snippets may now be
406 prefixed with "~", which indicates that they shall be masked
407 by whether the existing file or directly is currently
408 writable, readable or executable at all. Also, if specified,
409 the sgid/suid/sticky bits will be masked for all
412 * A new passive target unit "network-pre.target" has been
413 added which is useful for services that shall run before any
414 network is configured, for example firewall scripts.
416 * The "floppy" group that previously owned the /dev/fd*
417 devices is no longer used. The "disk" group is now used
418 instead. Distributions should probably deprecate usage of
421 Contributions from: Camilo Aguilar, Christian Hesse, Colin Ian
422 King, Cristian Rodríguez, Daniel Buch, Dave Reisner, David
423 Strauss, Denis Tikhomirov, John, Jonathan Liu, Kay Sievers,
424 Lennart Poettering, Mantas Mikulėnas, Mark Eichin, Ronny
425 Chevalier, Susant Sahani, Thomas Blume, Thomas Hindoe Paaboel
426 Andersen, Tom Gundersen, Umut Tezduyar Lindskog, Zbigniew
429 -- Berlin, 2014-06-11
433 * A new "systemd-timesyncd" daemon has been added for
434 synchronizing the system clock across the network. It
435 implements an SNTP client. In contrast to NTP
436 implementations such as chrony or the NTP reference server,
437 this only implements a client side, and does not bother with
438 the full NTP complexity, focusing only on querying time from
439 one remote server and synchronizing the local clock to
440 it. Unless you intend to serve NTP to networked clients or
441 want to connect to local hardware clocks, this simple NTP
442 client should be more than appropriate for most
443 installations. The daemon runs with minimal privileges, and
444 has been hooked up with networkd to only operate when
445 network connectivity is available. The daemon saves the
446 current clock to disk every time a new NTP sync has been
447 acquired, and uses this to possibly correct the system clock
448 early at bootup, in order to accommodate for systems that
449 lack an RTC such as the Raspberry Pi and embedded devices,
450 and to make sure that time monotonically progresses on these
451 systems, even if it is not always correct. To make use of
452 this daemon, a new system user and group "systemd-timesync"
453 needs to be created on installation of systemd.
455 * The queue "seqnum" interface of libudev has been disabled, as
456 it was generally incompatible with device namespacing as
457 sequence numbers of devices go "missing" if the devices are
458 part of a different namespace.
460 * "systemctl list-timers" and "systemctl list-sockets" gained
461 a --recursive switch for showing units of these types also
462 for all local containers, similar in style to the already
463 supported --recursive switch for "systemctl list-units".
465 * A new RebootArgument= setting has been added for service
466 units, which may be used to specify a kernel reboot argument
467 to use when triggering reboots with StartLimitAction=.
469 * A new FailureAction= setting has been added for service
470 units which may be used to specify an operation to trigger
471 when a service fails. This works similarly to
472 StartLimitAction=, but unlike it, controls what is done
473 immediately rather than only after several attempts to
474 restart the service in question.
476 * hostnamed got updated to also expose the kernel name,
477 release, and version on the bus. This is useful for
478 executing commands like hostnamectl with the -H switch.
479 systemd-analyze makes use of this to properly display
480 details when running non-locally.
482 * The bootchart tool can now show cgroup information in the
485 * The CFS CPU quota cgroup attribute is now exposed for
486 services. The new CPUQuota= switch has been added for this
487 which takes a percentage value. Setting this will have the
488 result that a service may never get more CPU time than the
489 specified percentage, even if the machine is otherwise idle.
491 * systemd-networkd learned IPIP and SIT tunnel support.
493 * LSB init scripts exposing a dependency on $network will now
494 get a dependency on network-online.target rather than simply
495 network.target. This should bring LSB handling closer to
496 what it was on SysV systems.
498 * A new fsck.repair= kernel option has been added to control
499 how fsck shall deal with unclean file systems at boot.
501 * The (.ini) configuration file parser will now silently
502 ignore sections whose name begins with "X-". This may be
503 used to maintain application-specific extension sections in unit
506 * machined gained a new API to query the IP addresses of
507 registered containers. "machinectl status" has been updated
508 to show these addresses in its output.
510 * A new call sd_uid_get_display() has been added to the
511 sd-login APIs for querying the "primary" session of a
512 user. The "primary" session of the user is elected from the
513 user's sessions and generally a graphical session is
514 preferred over a text one.
516 * A minimal systemd-resolved daemon has been added. It
517 currently simply acts as a companion to systemd-networkd and
518 manages resolv.conf based on per-interface DNS
519 configuration, possibly supplied via DHCP. In the long run
520 we hope to extend this into a local DNSSEC enabled DNS and
523 * The systemd-networkd-wait-online tool is now enabled by
524 default. It will delay network-online.target until a network
525 connection has been configured. The tool primarily integrates
526 with networkd, but will also make a best effort to make sense
527 of network configuration performed in some other way.
529 * Two new service options StartupCPUShares= and
530 StartupBlockIOWeight= have been added that work similarly to
531 CPUShares= and BlockIOWeight= however only apply during
532 system startup. This is useful to prioritize certain services
533 differently during bootup than during normal runtime.
535 * hostnamed has been changed to prefer the statically
536 configured hostname in /etc/hostname (unless set to
537 'localhost' or empty) over any dynamic one supplied by
538 dhcp. With this change, the rules for picking the hostname
539 match more closely the rules of other configuration settings
540 where the local administrator's configuration in /etc always
541 overrides any other settings.
543 Contributions fron: Ali H. Caliskan, Alison Chaiken, Bas van
544 den Berg, Brandon Philips, Cristian Rodríguez, Daniel Buch,
545 Dan Kilman, Dave Reisner, David Härdeman, David Herrmann,
546 David Strauss, Dimitris Spingos, Djalal Harouni, Eelco
547 Dolstra, Evan Nemerson, Florian Albrechtskirchinger, Greg
548 Kroah-Hartman, Harald Hoyer, Holger Hans Peter Freyther, Jan
549 Engelhardt, Jani Nikula, Jason St. John, Jeffrey Clark,
550 Jonathan Boulle, Kay Sievers, Lennart Poettering, Lukas
551 Nykryn, Lukasz Skalski, Łukasz Stelmach, Mantas Mikulėnas,
552 Marcel Holtmann, Martin Pitt, Matthew Monaco, Michael
553 Marineau, Michael Olbrich, Michal Sekletar, Mike Gilbert, Nis
554 Martensen, Patrik Flykt, Philip Lorenz, poma, Ray Strode,
555 Reyad Attiyat, Robert Milasan, Scott Thrasher, Stef Walter,
556 Steven Siloti, Susant Sahani, Tanu Kaskinen, Thomas Bächler,
557 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar
558 Lindskog, WaLyong Cho, Will Woods, Zbigniew
561 -- Beijing, 2014-05-28
565 * When restoring the screen brightness at boot, stay away from
566 the darkest setting or from the lowest 5% of the available
567 range, depending on which is the larger value of both. This
568 should effectively protect the user from rebooting into a
569 black screen, should the brightness have been set to minimum
572 * sd-login gained a new sd_machine_get_class() call to
573 determine the class ("vm" or "container") of a machine
574 registered with machined.
576 * sd-login gained new calls
577 sd_peer_get_{session,owner_uid,unit,user_unit,slice,machine_name}(),
578 to query the identity of the peer of a local AF_UNIX
579 connection. They operate similarly to their sd_pid_get_xyz()
582 * PID 1 will now maintain a system-wide system state engine
583 with the states "starting", "running", "degraded",
584 "maintenance", "stopping". These states are bound to system
585 startup, normal runtime, runtime with at least one failed
586 service, rescue/emergency mode and system shutdown. This
587 state is shown in the "systemctl status" output when no unit
588 name is passed. It is useful to determine system state, in
589 particularly when doing so for many systems or containers at
592 * A new command "list-machines" has been added to "systemctl"
593 that lists all local OS containers and shows their system
594 state (see above), if systemd runs inside of them.
596 * systemctl gained a new "-r" switch to recursively enumerate
597 units on all local containers, when used with the
598 "list-unit" command (which is the default one that is
599 executed when no parameters are specified).
601 * The GPT automatic partition discovery logic will now honour
602 two GPT partition flags: one may be set on a partition to
603 cause it to be mounted read-only, and the other may be set
604 on a partition to ignore it during automatic discovery.
606 * Two new GPT type UUIDs have been added for automatic root
607 partition discovery, for 32-bit and 64-bit ARM. This is not
608 particularly useful for discovering the root directory on
609 these architectures during bare-metal boots (since UEFI is
610 not common there), but still very useful to allow booting of
611 ARM disk images in nspawn with the -i option.
613 * MAC addresses of interfaces created with nspawn's
614 --network-interface= switch will now be generated from the
615 machine name, and thus be stable between multiple invocations
618 * logind will now automatically remove all IPC objects owned
619 by a user if she or he fully logs out. This makes sure that
620 users who are logged out cannot continue to consume IPC
621 resources. This covers SysV memory, semaphores and message
622 queues as well as POSIX shared memory and message
623 queues. Traditionally, SysV and POSIX IPC had no life-cycle
624 limits. With this functionality, that is corrected. This may
625 be turned off by using the RemoveIPC= switch of logind.conf.
627 * The systemd-machine-id-setup and tmpfiles tools gained a
628 --root= switch to operate on a specific root directory,
631 * journald can now forward logged messages to the TTYs of all
632 logged in users ("wall"). This is the default for all
633 emergency messages now.
635 * A new tool systemd-journal-remote has been added to stream
636 journal log messages across the network.
638 * /sys/fs/cgroup/ is now mounted read-only after all cgroup
639 controller trees are mounted into it. Note that the
640 directories mounted beneath it are not read-only. This is a
641 security measure and is particularly useful because glibc
642 actually includes a search logic to pick any tmpfs it can
643 find to implement shm_open() if /dev/shm is not available
644 (which it might very well be in namespaced setups).
646 * machinectl gained a new "poweroff" command to cleanly power
647 down a local OS container.
649 * The PrivateDevices= unit file setting will now also drop the
650 CAP_MKNOD capability from the capability bound set, and
651 imply DevicePolicy=closed.
653 * PrivateDevices=, PrivateNetwork= and PrivateTmp= is now used
654 comprehensively on all long-running systemd services where
657 * systemd-udevd will now run in a disassociated mount
658 namespace. To mount directories from udev rules, make sure to
659 pull in mount units via SYSTEMD_WANTS properties.
661 * The kdbus support gained support for uploading policy into
662 the kernel. sd-bus gained support for creating "monitoring"
663 connections that can eavesdrop into all bus communication
664 for debugging purposes.
666 * Timestamps may now be specified in seconds since the UNIX
667 epoch Jan 1st, 1970 by specifying "@" followed by the value
670 * Native tcpwrap support in systemd has been removed. tcpwrap
671 is old code, not really maintained anymore and has serious
672 shortcomings, and better options such as firewalls
673 exist. For setups that require tcpwrap usage, please
674 consider invoking your socket-activated service via tcpd,
675 like on traditional inetd.
677 * A new system.conf configuration option
678 DefaultTimerAccuracySec= has been added that controls the
679 default AccuracySec= setting of .timer units.
681 * Timer units gained a new WakeSystem= switch. If enabled,
682 timers configured this way will cause the system to resume
683 from system suspend (if the system supports that, which most
686 * Timer units gained a new Persistent= switch. If enabled,
687 timers configured this way will save to disk when they have
688 been last triggered. This information is then used on next
689 reboot to possible execute overdue timer events, that
690 could not take place because the system was powered off.
691 This enables simple anacron-like behaviour for timer units.
693 * systemctl's "list-timers" will now also list the time a
694 timer unit was last triggered in addition to the next time
695 it will be triggered.
697 * systemd-networkd will now assign predictable IPv4LL
698 addresses to its local interfaces.
700 Contributions from: Brandon Philips, Daniel Buch, Daniel Mack,
701 Dave Reisner, David Herrmann, Gerd Hoffmann, Greg
702 Kroah-Hartman, Hendrik Brueckner, Jason St. John, Josh
703 Triplett, Kay Sievers, Lennart Poettering, Marc-Antoine
704 Perennou, Michael Marineau, Michael Olbrich, Miklos Vajna,
705 Patrik Flykt, poma, Sebastian Thorarensen, Thomas Bächler,
706 Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom Gundersen,
707 Umut Tezduyar Lindskog, Wieland Hoffmann, Zbigniew
710 -- Berlin, 2014-03-25
714 * A new unit file setting RestrictAddressFamilies= has been
715 added to restrict which socket address families unit
716 processes gain access to. This takes address family names
717 like "AF_INET" or "AF_UNIX", and is useful to minimize the
718 attack surface of services via exotic protocol stacks. This
719 is built on seccomp system call filters.
721 * Two new unit file settings RuntimeDirectory= and
722 RuntimeDirectoryMode= have been added that may be used to
723 manage a per-daemon runtime directories below /run. This is
724 an alternative for setting up directory permissions with
725 tmpfiles snippets, and has the advantage that the runtime
726 directory's lifetime is bound to the daemon runtime and that
727 the daemon starts up with an empty directory each time. This
728 is particularly useful when writing services that drop
729 privileges using the User= or Group= setting.
731 * The DeviceAllow= unit setting now supports globbing for
732 matching against device group names.
734 * The systemd configuration file system.conf gained new
735 settings DefaultCPUAccounting=, DefaultBlockIOAccounting=,
736 DefaultMemoryAccounting= to globally turn on/off accounting
737 for specific resources (cgroups) for all units. These
738 settings may still be overridden individually in each unit
741 * systemd-gpt-auto-generator is now able to discover /srv and
742 root partitions in addition to /home and swap partitions. It
743 also supports LUKS-encrypted partitions now. With this in
744 place, automatic discovery of partitions to mount following
745 the Discoverable Partitions Specification
746 (http://www.freedesktop.org/wiki/Specifications/DiscoverablePartitionsSpec)
747 is now a lot more complete. This allows booting without
748 /etc/fstab and without root= on the kernel command line on
749 systems prepared appropriately.
751 * systemd-nspawn gained a new --image= switch which allows
752 booting up disk images and Linux installations on any block
753 device that follow the Discoverable Partitions Specification
754 (see above). This means that installations made with
755 appropriately updated installers may now be started and
756 deployed using container managers, completely
757 unmodified. (We hope that libvirt-lxc will add support for
758 this feature soon, too.)
760 * systemd-nspawn gained a new --network-macvlan= setting to
761 set up a private macvlan interface for the
762 container. Similarly, systemd-networkd gained a new
763 Kind=macvlan setting in .netdev files.
765 * systemd-networkd now supports configuring local addresses
768 * A new tool systemd-network-wait-online has been added to
769 synchronously wait for network connectivity using
772 * The sd-bus.h bus API gained a new sd_bus_track object for
773 tracking the life-cycle of bus peers. Note that sd-bus.h is
774 still not a public API though (unless you specify
775 --enable-kdbus on the configure command line, which however
776 voids your warranty and you get no API stability guarantee).
778 * The $XDG_RUNTIME_DIR runtime directories for each user are
779 now individual tmpfs instances, which has the benefit of
780 introducing separate pools for each user, with individual
781 size limits, and thus making sure that unprivileged clients
782 can no longer negatively impact the system or other users by
783 filling up their $XDG_RUNTIME_DIR. A new logind.conf setting
784 RuntimeDirectorySize= has been introduced that allows
785 controlling the default size limit for all users. It
786 defaults to 10% of the available physical memory. This is no
787 replacement for quotas on tmpfs though (which the kernel
788 still does not support), as /dev/shm and /tmp are still
789 shared resources used by both the system and unprivileged
792 * logind will now automatically turn off automatic suspending
793 on laptop lid close when more than one display is
794 connected. This was previously expected to be implemented
795 individually in desktop environments (such as GNOME),
796 however has been added to logind now, in order to fix a
797 boot-time race where a desktop environment might not have
798 been started yet and thus not been able to take an inhibitor
799 lock at the time where logind already suspends the system
802 * logind will now wait at least 30s after each system
803 suspend/resume cycle, and 3min after system boot before
804 suspending the system due to a closed laptop lid. This
805 should give USB docking stations and similar enough time to
806 be probed and configured after system resume and boot in
807 order to then act as suspend blocker.
809 * systemd-run gained a new --property= setting which allows
810 initialization of resource control properties (and others)
811 for the created scope or service unit. Example: "systemd-run
812 --property=BlockIOWeight=10 updatedb" may be used to run
813 updatedb at a low block IO scheduling weight.
815 * systemd-run's --uid=, --gid=, --setenv=, --setenv= switches
816 now also work in --scope mode.
818 * When systemd is compiled with kdbus support, basic support
819 for enforced policies is now in place. (Note that enabling
820 kdbus still voids your warranty and no API compatibility
823 Contributions from: Andrey Borzenkov, Ansgar Burchardt, Armin
824 K., Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
825 Harald Hoyer, Henrik Grindal Bakken, Jasper St. Pierre, Kay
826 Sievers, Kieran Clancy, Lennart Poettering, Lukas Nykryn,
827 Mantas Mikulėnas, Marcel Holtmann, Mark Oteiza, Martin Pitt,
828 Mike Gilbert, Peter Rajnoha, poma, Samuli Suominen, Stef
829 Walter, Susant Sahani, Tero Roponen, Thomas Andersen, Thomas
830 Bächler, Thomas Hindoe Paaboel Andersen, Tomasz Torcz, Tom
831 Gundersen, Umut Tezduyar Lindskog, Uoti Urpala, Zachary Cook,
832 Zbigniew Jędrzejewski-Szmek
834 -- Berlin, 2014-03-12
838 * systemd will now relabel /dev after loading the SMACK policy
839 according to SMACK rules.
841 * A new unit file option AppArmorProfile= has been added to
842 set the AppArmor profile for the processes of a unit.
844 * A new condition check ConditionArchitecture= has been added
845 to conditionalize units based on the system architecture, as
846 reported by uname()'s "machine" field.
848 * systemd-networkd now supports matching on the system
849 virtualization, architecture, kernel command line, host name
852 * logind is now a lot more aggressive when suspending the
853 machine due to a closed laptop lid. Instead of acting only
854 on the lid close action, it will continuously watch the lid
855 status and act on it. This is useful for laptops where the
856 power button is on the outside of the chassis so that it can
857 be reached without opening the lid (such as the Lenovo
858 Yoga). On those machines, logind will now immediately
859 re-suspend the machine if the power button has been
860 accidentally pressed while the laptop was suspended and in a
863 * logind will now watch SW_DOCK switches and inhibit reaction
864 to the lid switch if it is pressed. This means that logind
865 will not suspend the machine anymore if the lid is closed
866 and the system is docked, if the laptop supports SW_DOCK
867 notifications via the input layer. Note that ACPI docking
868 stations do not generate this currently. Also note that this
869 logic is usually not fully sufficient and Desktop
870 Environments should take a lid switch inhibitor lock when an
871 external display is connected, as systemd will not watch
874 * nspawn will now make use of the devices cgroup controller by
875 default, and only permit creation of and access to the usual
876 API device nodes like /dev/null or /dev/random, as well as
877 access to (but not creation of) the pty devices.
879 * We will now ship a default .network file for
880 systemd-networkd that automatically configures DHCP for
881 network interfaces created by nspawn's --network-veth or
882 --network-bridge= switches.
884 * systemd will now understand the usual M, K, G, T suffixes
885 according to SI conventions (i.e. to the base 1000) when
886 referring to throughput and hardware metrics. It will stay
887 with IEC conventions (i.e. to the base 1024) for software
888 metrics, according to what is customary according to
889 Wikipedia. We explicitly document which base applies for
890 each configuration option.
892 * The DeviceAllow= setting in unit files now supports a syntax
893 to whitelist an entire group of devices node majors at once,
894 based on the /proc/devices listing. For example, with the
895 string "char-pts", it is now possible to whitelist all
896 current and future pseudo-TTYs at once.
898 * sd-event learned a new "post" event source. Event sources of
899 this type are triggered by the dispatching of any event
900 source of a type that is not "post". This is useful for
901 implementing clean-up and check event sources that are
902 triggered by other work being done in the program.
904 * systemd-networkd is no longer statically enabled, but uses
905 the usual [Install] sections so that it can be
906 enabled/disabled using systemctl. It still is enabled by
909 * When creating a veth interface pair with systemd-nspawn, the
910 host side will now be prefixed with "vb-" if
911 --network-bridge= is used, and with "ve-" if --network-veth
912 is used. This way, it is easy to distinguish these cases on
913 the host, for example to apply different configuration to
914 them with systemd-networkd.
916 * The compatibility libraries for libsystemd-journal.so,
917 libsystem-id128.so, libsystemd-login.so and
918 libsystemd-daemon.so do not make use of IFUNC
919 anymore. Instead, we now build libsystemd.so multiple times
920 under these alternative names. This means that the footprint
921 is drastically increased, but given that these are
922 transitional compatibility libraries, this should not matter
923 much. This change has been made necessary to support the ARM
924 platform for these compatibility libraries, as the ARM
925 toolchain is not really at the same level as the toolchain
926 for other architectures like x86 and does not support
927 IFUNC. Please make sure to use --enable-compat-libs only
928 during a transitional period!
930 Contributions from: Andreas Fuchs, Armin K., Colin Walters,
931 Daniel Mack, Dave Reisner, David Herrmann, Djalal Harouni,
932 Holger Schurig, Jason A. Donenfeld, Jason St. John, Jasper
933 St. Pierre, Kay Sievers, Lennart Poettering, Łukasz Stelmach,
934 Marcel Holtmann, Michael Scherer, Michal Sekletar, Mike
935 Gilbert, Samuli Suominen, Thomas Bächler, Thomas Hindoe
936 Paaboel Andersen, Tom Gundersen, Umut Tezduyar Lindskog,
937 Zbigniew Jędrzejewski-Szmek
939 -- Berlin, 2014-02-24
943 * A new component "systemd-networkd" has been added that can
944 be used to configure local network interfaces statically or
945 via DHCP. It is capable of bringing up bridges, VLANs, and
946 bonding. Currently, no hook-ups for interactive network
947 configuration are provided. Use this for your initrd,
948 container, embedded, or server setup if you need a simple,
949 yet powerful, network configuration solution. This
950 configuration subsystem is quite nifty, as it allows wildcard
951 hotplug matching in interfaces. For example, with a single
952 configuration snippet, you can configure that all Ethernet
953 interfaces showing up are automatically added to a bridge,
954 or similar. It supports link-sensing and more.
956 * A new tool "systemd-socket-proxyd" has been added which can
957 act as a bidirectional proxy for TCP sockets. This is
958 useful for adding socket activation support to services that
959 do not actually support socket activation, including virtual
960 machines and the like.
962 * Add a new tool to save/restore rfkill state on
965 * Save/restore state of keyboard backlights in addition to
966 display backlights on shutdown/boot.
968 * udev learned a new SECLABEL{} construct to label device
969 nodes with a specific security label when they appear. For
970 now, only SECLABEL{selinux} is supported, but the syntax is
971 prepared for additional security frameworks.
973 * udev gained a new scheme to configure link-level attributes
974 from files in /etc/systemd/network/*.link. These files can
975 match against MAC address, device path, driver name and type,
976 and will apply attributes like the naming policy, link speed,
977 MTU, duplex settings, Wake-on-LAN settings, MAC address, MAC
978 address assignment policy (randomized, ...).
980 * The configuration of network interface naming rules for
981 "permanent interface names" has changed: a new NamePolicy=
982 setting in the [Link] section of .link files determines the
983 priority of possible naming schemes (onboard, slot, mac,
984 path). The default value of this setting is determined by
985 /usr/lib/net/links/99-default.link. Old
986 80-net-name-slot.rules udev configuration file has been
987 removed, so local configuration overriding this file should
988 be adapated to override 99-default.link instead.
990 * When the User= switch is used in a unit file, also
991 initialize $SHELL= based on the user database entry.
993 * systemd no longer depends on libdbus. All communication is
994 now done with sd-bus, systemd's low-level bus library
997 * kdbus support has been added to PID 1 itself. When kdbus is
998 enabled, this causes PID 1 to set up the system bus and
999 enable support for a new ".busname" unit type that
1000 encapsulates bus name activation on kdbus. It works a little
1001 bit like ".socket" units, except for bus names. A new
1002 generator has been added that converts classic dbus1 service
1003 activation files automatically into native systemd .busname
1006 * sd-bus: add a light-weight vtable implementation that allows
1007 defining objects on the bus with a simple static const
1008 vtable array of its methods, signals and properties.
1010 * systemd will not generate or install static dbus
1011 introspection data anymore to /usr/share/dbus-1/interfaces,
1012 as the precise format of these files is unclear, and
1013 nothing makes use of it.
1015 * A proxy daemon is now provided to proxy clients connecting
1016 via classic D-Bus AF_UNIX sockets to kdbus, to provide full
1017 compatibility with classic D-Bus.
1019 * A bus driver implementation has been added that supports the
1020 classic D-Bus bus driver calls on kdbus, also for
1021 compatibility purposes.
1023 * A new API "sd-event.h" has been added that implements a
1024 minimal event loop API built around epoll. It provides a
1025 couple of features that direct epoll usage is lacking:
1026 prioritization of events, scales to large numbers of timer
1027 events, per-event timer slack (accuracy), system-wide
1028 coalescing of timer events, exit handlers, watchdog
1029 supervision support using systemd's sd_notify() API, child
1032 * A new API "sd-rntl.h" has been added that provides an API
1033 around the route netlink interface of the kernel, similar in
1034 style to "sd-bus.h".
1036 * A new API "sd-dhcp-client.h" has been added that provides a
1037 small DHCPv4 client-side implementation. This is used by
1040 * There is a new kernel command line option
1041 "systemd.restore_state=0|1". When set to "0", none of the
1042 systemd tools will restore saved runtime state to hardware
1043 devices. More specifically, the rfkill and backlight states
1046 * The FsckPassNo= compatibility option in mount/service units
1047 has been removed. The fstab generator will now add the
1048 necessary dependencies automatically, and does not require
1049 PID1's support for that anymore.
1051 * journalctl gained a new switch, --list-boots, that lists
1052 recent boots with their times and boot IDs.
1054 * The various tools like systemctl, loginctl, timedatectl,
1055 busctl, systemd-run, ... have gained a new switch "-M" to
1056 connect to a specific, local OS container (as direct
1057 connection, without requiring SSH). This works on any
1058 container that is registered with machined, such as those
1059 created by libvirt-lxc or nspawn.
1061 * systemd-run and systemd-analyze also gained support for "-H"
1062 to connect to remote hosts via SSH. This is particularly
1063 useful for systemd-run because it enables queuing of jobs
1064 onto remote systems.
1066 * machinectl gained a new command "login" to open a getty
1067 login in any local container. This works with any container
1068 that is registered with machined (such as those created by
1069 libvirt-lxc or nspawn), and which runs systemd inside.
1071 * machinectl gained a new "reboot" command that may be used to
1072 trigger a reboot on a specific container that is registered
1073 with machined. This works on any container that runs an init
1074 system of some kind.
1076 * systemctl gained a new "list-timers" command to print a nice
1077 listing of installed timer units with the times they elapse
1080 * Alternative reboot() parameters may now be specified on the
1081 "systemctl reboot" command line and are passed to the
1082 reboot() system call.
1084 * systemctl gained a new --job-mode= switch to configure the
1085 mode to queue a job with. This is a more generic version of
1086 --fail, --irreversible, and --ignore-dependencies, which are
1087 still available but not advertised anymore.
1089 * /etc/systemd/system.conf gained new settings to configure
1090 various default timeouts of units, as well as the default
1091 start limit interval and burst. These may still be overridden
1094 * PID1 will now export on the bus profile data of the security
1095 policy upload process (such as the SELinux policy upload to
1098 * journald: when forwarding logs to the console, include
1099 timestamps (following the setting in
1100 /sys/module/printk/parameters/time).
1102 * OnCalendar= in timer units now understands the special
1103 strings "yearly" and "annually". (Both are equivalent)
1105 * The accuracy of timer units is now configurable with the new
1106 AccuracySec= setting. It defaults to 1min.
1108 * A new dependency type JoinsNamespaceOf= has been added that
1109 allows running two services within the same /tmp and network
1110 namespace, if PrivateNetwork= or PrivateTmp= are used.
1112 * A new command "cat" has been added to systemctl. It outputs
1113 the original unit file of a unit, and concatenates the
1114 contents of additional "drop-in" unit file snippets, so that
1115 the full configuration is shown.
1117 * systemctl now supports globbing on the various "list-xyz"
1118 commands, like "list-units" or "list-sockets", as well as on
1119 those commands which take multiple unit names.
1121 * journalctl's --unit= switch gained support for globbing.
1123 * All systemd daemons now make use of the watchdog logic so
1124 that systemd automatically notices when they hang.
1126 * If the $container_ttys environment variable is set,
1127 getty-generator will automatically spawn a getty for each
1128 listed tty. This is useful for container managers to request
1129 login gettys to be spawned on as many ttys as needed.
1131 * %h, %s, %U specifier support is not available anymore when
1132 used in unit files for PID 1. This is because NSS calls are
1133 not safe from PID 1. They stay available for --user
1134 instances of systemd, and as special case for the root user.
1136 * loginctl gained a new "--no-legend" switch to turn off output
1139 * The "sd-login.h" API gained three new calls:
1140 sd_session_is_remote(), sd_session_get_remote_user(),
1141 sd_session_get_remote_host() to query information about
1144 * The udev hardware database now also carries vendor/product
1145 information of SDIO devices.
1147 * The "sd-daemon.h" API gained a new sd_watchdog_enabled() to
1148 determine whether watchdog notifications are requested by
1151 * Socket-activated per-connection services now include a
1152 short description of the connection parameters in the
1155 * tmpfiles gained a new "--boot" option. When this is not used,
1156 only lines where the command character is not suffixed with
1157 "!" are executed. When this option is specified, those
1158 options are executed too. This partitions tmpfiles
1159 directives into those that can be safely executed at any
1160 time, and those which should be run only at boot (for
1161 example, a line that creates /run/nologin).
1163 * A new API "sd-resolve.h" has been added which provides a simple
1164 asynchronous wrapper around glibc NSS host name resolution
1165 calls, such as getaddrinfo(). In contrast to glibc's
1166 getaddrinfo_a(), it does not use signals. In contrast to most
1167 other asynchronous name resolution libraries, this one does
1168 not reimplement DNS, but reuses NSS, so that alternate
1169 host name resolution systems continue to work, such as mDNS,
1170 LDAP, etc. This API is based on libasyncns, but it has been
1171 cleaned up for inclusion in systemd.
1173 * The APIs "sd-journal.h", "sd-login.h", "sd-id128.h",
1174 "sd-daemon.h" are no longer found in individual libraries
1175 libsystemd-journal.so, libsystemd-login.so,
1176 libsystemd-id128.so, libsystemd-daemon.so. Instead, we have
1177 merged them into a single library, libsystemd.so, which
1178 provides all symbols. The reason for this is cyclic
1179 dependencies, as these libraries tend to use each other's
1180 symbols. So far, we have managed to workaround that by linking
1181 a copy of a good part of our code into each of these
1182 libraries again and again, which, however, makes certain
1183 things hard to do, like sharing static variables. Also, it
1184 substantially increases footprint. With this change, there
1185 is only one library for the basic APIs systemd
1186 provides. Also, "sd-bus.h", "sd-memfd.h", "sd-event.h",
1187 "sd-rtnl.h", "sd-resolve.h", "sd-utf8.h" are found in this
1188 library as well, however are subject to the --enable-kdbus
1189 switch (see below). Note that "sd-dhcp-client.h" is not part
1190 of this library (this is because it only consumes, never
1191 provides, services of/to other APIs). To make the transition
1192 easy from the separate libraries to the unified one, we
1193 provide the --enable-compat-libs compile-time switch which
1194 will generate stub libraries that are compatible with the
1195 old ones but redirect all calls to the new one.
1197 * All of the kdbus logic and the new APIs "sd-bus.h",
1198 "sd-memfd.h", "sd-event.h", "sd-rtnl.h", "sd-resolve.h",
1199 and "sd-utf8.h" are compile-time optional via the
1200 "--enable-kdbus" switch, and they are not compiled in by
1201 default. To make use of kdbus, you have to explicitly enable
1202 the switch. Note however, that neither the kernel nor the
1203 userspace API for all of this is considered stable yet. We
1204 want to maintain the freedom to still change the APIs for
1205 now. By specifying this build-time switch, you acknowledge
1206 that you are aware of the instability of the current
1209 * Also, note that while kdbus is pretty much complete,
1210 it lacks one thing: proper policy support. This means you
1211 can build a fully working system with all features; however,
1212 it will be highly insecure. Policy support will be added in
1213 one of the next releases, at the same time that we will
1214 declare the APIs stable.
1216 * When the kernel command-line argument "kdbus" is specified,
1217 systemd will automatically load the kdbus.ko kernel module. At
1218 this stage of development, it is only useful for testing kdbus
1219 and should not be used in production. Note: if "--enable-kdbus"
1220 is specified, and the kdbus.ko kernel module is available, and
1221 "kdbus" is added to the kernel command line, the entire system
1222 runs with kdbus instead of dbus-daemon, with the above mentioned
1223 problem of missing the system policy enforcement. Also a future
1224 version of kdbus.ko or a newer systemd will not be compatible with
1225 each other, and will unlikely be able to boot the machine if only
1226 one of them is updated.
1228 * systemctl gained a new "import-environment" command which
1229 uploads the caller's environment (or parts thereof) into the
1230 service manager so that it is inherited by services started
1231 by the manager. This is useful to upload variables like
1232 $DISPLAY into the user service manager.
1234 * A new PrivateDevices= switch has been added to service units
1235 which allows running a service with a namespaced /dev
1236 directory that does not contain any device nodes for
1237 physical devices. More specifically, it only includes devices
1238 such as /dev/null, /dev/urandom, and /dev/zero which are API
1241 * logind has been extended to support behaviour like VT
1242 switching on seats that do not support a VT. This makes
1243 multi-session available on seats that are not the first seat
1244 (seat0), and on systems where kernel support for VTs has
1245 been disabled at compile-time.
1247 * If a process holds a delay lock for system sleep or shutdown
1248 and fails to release it in time, we will now log its
1249 identity. This makes it easier to identify processes that
1250 cause slow suspends or power-offs.
1252 * When parsing /etc/crypttab, support for a new key-slot=
1253 option as supported by Debian is added. It allows indicating
1254 which LUKS slot to use on disk, speeding up key loading.
1256 * The sd_journald_sendv() API call has been checked and
1257 officially declared to be async-signal-safe so that it may
1258 be invoked from signal handlers for logging purposes.
1260 * Boot-time status output is now enabled automatically after a
1261 short timeout if boot does not progress, in order to give
1262 the user an indication what she or he is waiting for.
1264 * The boot-time output has been improved to show how much time
1265 remains until jobs expire.
1267 * The KillMode= switch in service units gained a new possible
1268 value "mixed". If set, and the unit is shut down, then the
1269 initial SIGTERM signal is sent only to the main daemon
1270 process, while the following SIGKILL signal is sent to
1271 all remaining processes of the service.
1273 * When a scope unit is registered, a new property "Controller"
1274 may be set. If set to a valid bus name, systemd will send a
1275 RequestStop() signal to this name when it would like to shut
1276 down the scope. This may be used to hook manager logic into
1277 the shutdown logic of scope units. Also, scope units may now
1278 be put in a special "abandoned" state, in which case the
1279 manager process which created them takes no further
1280 responsibilities for it.
1282 * When reading unit files, systemd will now verify
1283 the access mode of these files, and warn about certain
1284 suspicious combinations. This has been added to make it
1285 easier to track down packaging bugs where unit files are
1286 marked executable or world-writable.
1288 * systemd-nspawn gained a new "--setenv=" switch to set
1289 container-wide environment variables. The similar option in
1290 systemd-activate was renamed from "--environment=" to
1291 "--setenv=" for consistency.
1293 * systemd-nspawn has been updated to create a new kdbus domain
1294 for each container that is invoked, thus allowing each
1295 container to have its own set of system and user buses,
1296 independent of the host.
1298 * systemd-nspawn gained a new --drop-capability= switch to run
1299 the container with less capabilities than the default. Both
1300 --drop-capability= and --capability= now take the special
1301 string "all" for dropping or keeping all capabilities.
1303 * systemd-nspawn gained new switches for executing containers
1304 with specific SELinux labels set.
1306 * systemd-nspawn gained a new --quiet switch to not generate
1307 any additional output but the container's own console
1310 * systemd-nspawn gained a new --share-system switch to run a
1311 container without PID namespacing enabled.
1313 * systemd-nspawn gained a new --register= switch to control
1314 whether the container is registered with systemd-machined or
1315 not. This is useful for containers that do not run full
1316 OS images, but only specific apps.
1318 * systemd-nspawn gained a new --keep-unit which may be used
1319 when invoked as the only program from a service unit, and
1320 results in registration of the unit service itself in
1321 systemd-machined, instead of a newly opened scope unit.
1323 * systemd-nspawn gained a new --network-interface= switch for
1324 moving arbitrary interfaces to the container. The new
1325 --network-veth switch creates a virtual Ethernet connection
1326 between host and container. The new --network-bridge=
1327 switch then allows assigning the host side of this virtual
1328 Ethernet connection to a bridge device.
1330 * systemd-nspawn gained a new --personality= switch for
1331 setting the kernel personality for the container. This is
1332 useful when running a 32-bit container on a 64-bit host. A
1333 similar option Personality= is now also available for service
1336 * logind will now also track a "Desktop" identifier for each
1337 session which encodes the desktop environment of it. This is
1338 useful for desktop environments that want to identify
1339 multiple running sessions of itself easily.
1341 * A new SELinuxContext= setting for service units has been
1342 added that allows setting a specific SELinux execution
1343 context for a service.
1345 * Most systemd client tools will now honour $SYSTEMD_LESS for
1346 settings of the "less" pager. By default, these tools will
1347 override $LESS to allow certain operations to work, such as
1348 jump-to-the-end. With $SYSTEMD_LESS, it is possible to
1349 influence this logic.
1351 * systemd's "seccomp" hook-up has been changed to make use of
1352 the libseccomp library instead of using its own
1353 implementation. This has benefits for portability among
1356 * For usage together with SystemCallFilter=, a new
1357 SystemCallErrorNumber= setting has been introduced that
1358 allows configuration of a system error number to be returned
1359 on filtered system calls, instead of immediately killing the
1360 process. Also, SystemCallArchitectures= has been added to
1361 limit access to system calls of a particular architecture
1362 (in order to turn off support for unused secondary
1363 architectures). There is also a global
1364 SystemCallArchitectures= setting in system.conf now to turn
1365 off support for non-native system calls system-wide.
1367 * systemd requires a kernel with a working name_to_handle_at(),
1368 please see the kernel config requirements in the README file.
1370 Contributions from: Adam Williamson, Alex Jia, Anatol Pomozov,
1371 Ansgar Burchardt, AppleBloom, Auke Kok, Bastien Nocera,
1372 Chengwei Yang, Christian Seiler, Colin Guthrie, Colin Walters,
1373 Cristian Rodríguez, Daniel Buch, Daniele Medri, Daniel J
1374 Walsh, Daniel Mack, Dan McGee, Dave Reisner, David Coppa,
1375 David Herrmann, David Strauss, Djalal Harouni, Dmitry Pisklov,
1376 Elia Pinto, Florian Weimer, George McCollister, Goffredo
1377 Baroncelli, Greg Kroah-Hartman, Hendrik Brueckner, Igor
1378 Zhbanov, Jan Engelhardt, Jan Janssen, Jason A. Donenfeld,
1379 Jason St. John, Jasper St. Pierre, Jóhann B. Guðmundsson, Jose
1380 Ignacio Naranjo, Karel Zak, Kay Sievers, Kristian Høgsberg,
1381 Lennart Poettering, Lubomir Rintel, Lukas Nykryn, Lukasz
1382 Skalski, Łukasz Stelmach, Luke Shumaker, Mantas Mikulėnas,
1383 Marc-Antoine Perennou, Marcel Holtmann, Marcos Felipe Rasia de
1384 Mello, Marko Myllynen, Martin Pitt, Matthew Monaco, Michael
1385 Marineau, Michael Scherer, Michał Górny, Michal Sekletar,
1386 Michele Curti, Oleksii Shevchuk, Olivier Brunel, Patrik Flykt,
1387 Pavel Holica, Raudi, Richard Marko, Ronny Chevalier, Sébastien
1388 Luttringer, Sergey Ptashnick, Shawn Landden, Simon Peeters,
1389 Stefan Beller, Susant Sahani, Sylvain Plantefeve, Sylvia Else,
1390 Tero Roponen, Thomas Bächler, Thomas Hindoe Paaboel Andersen,
1391 Tom Gundersen, Umut Tezduyar Lindskog, Unai Uribarri, Václav
1392 Pavlín, Vincent Batts, WaLyong Cho, William Giokas, Yang
1393 Zhiyong, Yin Kangkai, Yuxuan Shui, Zbigniew Jędrzejewski-Szmek
1395 -- Berlin, 2014-02-20
1399 * logind has gained support for facilitating privileged input
1400 and drm device access for unprivileged clients. This work is
1401 useful to allow Wayland display servers (and similar
1402 programs, such as kmscon) to run under the user's ID and
1403 access input and drm devices which are normally
1404 protected. When this is used (and the kernel is new enough)
1405 logind will "mute" IO on the file descriptors passed to
1406 Wayland as long as it is in the background and "unmute" it
1407 if it returns into the foreground. This allows secure
1408 session switching without allowing background sessions to
1409 eavesdrop on input and display data. This also introduces
1410 session switching support if VT support is turned off in the
1411 kernel, and on seats that are not seat0.
1413 * A new kernel command line option luks.options= is understood
1414 now which allows specifiying LUKS options for usage for LUKS
1415 encrypted partitions specified with luks.uuid=.
1417 * tmpfiles.d(5) snippets may now use specifier expansion in
1418 path names. More specifically %m, %b, %H, %v, are now
1419 replaced by the local machine id, boot id, hostname, and
1420 kernel version number.
1422 * A new tmpfiles.d(5) command "m" has been introduced which
1423 may be used to change the owner/group/access mode of a file
1424 or directory if it exists, but do nothing if it does not.
1426 * This release removes high-level support for the
1427 MemorySoftLimit= cgroup setting. The underlying kernel
1428 cgroup attribute memory.soft_limit= is currently badly
1429 designed and likely to be removed from the kernel API in its
1430 current form, hence we should not expose it for now.
1432 * The memory.use_hierarchy cgroup attribute is now enabled for
1433 all cgroups systemd creates in the memory cgroup
1434 hierarchy. This option is likely to be come the built-in
1435 default in the kernel anyway, and the non-hierarchial mode
1436 never made much sense in the intrinsically hierarchial
1439 * A new field _SYSTEMD_SLICE= is logged along with all journal
1440 messages containing the slice a message was generated
1441 from. This is useful to allow easy per-customer filtering of
1442 logs among other things.
1444 * systemd-journald will no longer adjust the group of journal
1445 files it creates to the "systemd-journal" group. Instead we
1446 rely on the journal directory to be owned by the
1447 "systemd-journal" group, and its setgid bit set, so that the
1448 kernel file system layer will automatically enforce that
1449 journal files inherit this group assignment. The reason for
1450 this change is that we cannot allow NSS look-ups from
1451 journald which would be necessary to resolve
1452 "systemd-journal" to a numeric GID, because this might
1453 create deadlocks if NSS involves synchronous queries to
1454 other daemons (such as nscd, or sssd) which in turn are
1455 logging clients of journald and might block on it, which
1456 would then dead lock. A tmpfiles.d(5) snippet included in
1457 systemd will make sure the setgid bit and group are
1458 properly set on the journal directory if it exists on every
1459 boot. However, we recommend adjusting it manually after
1460 upgrades too (or from RPM scriptlets), so that the change is
1461 not delayed until next reboot.
1463 * Backlight and random seed files in /var/lib/ have moved into
1464 the /var/lib/systemd/ directory, in order to centralize all
1465 systemd generated files in one directory.
1467 * Boot time performance measurements (as displayed by
1468 "systemd-analyze" for example) will now read ACPI 5.0 FPDT
1469 performance information if that's available to determine how
1470 much time BIOS and boot loader initialization required. With
1471 a sufficiently new BIOS you hence no longer need to boot
1472 with Gummiboot to get access to such information.
1474 Contributions from: Andrey Borzenkov, Chen Jie, Colin Walters,
1475 Cristian Rodríguez, Dave Reisner, David Herrmann, David
1476 Mackey, David Strauss, Eelco Dolstra, Evan Callicoat, Gao
1477 feng, Harald Hoyer, Jimmie Tauriainen, Kay Sievers, Lennart
1478 Poettering, Lukas Nykryn, Mantas Mikulėnas, Martin Pitt,
1479 Michael Scherer, Michał Górny, Mike Gilbert, Patrick McCarty,
1480 Sebastian Ott, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1482 -- Berlin, 2013-10-02
1486 * The Restart= option for services now understands a new
1487 on-watchdog setting, which will restart the service
1488 automatically if the service stops sending out watchdog keep
1489 alive messages (as configured with WatchdogSec=).
1491 * The getty generator (which is responsible for bringing up a
1492 getty on configured serial consoles) will no longer only
1493 start a getty on the primary kernel console but on all
1494 others, too. This makes the order in which console= is
1495 specified on the kernel command line less important.
1497 * libsystemd-logind gained a new sd_session_get_vt() call to
1498 retrieve the VT number of a session.
1500 * If the option "tries=0" is set for an entry of /etc/crypttab
1501 its passphrase is queried indefinitely instead of any
1502 maximum number of tries.
1504 * If a service with a configure PID file terminates its PID
1505 file will now be removed automatically if it still exists
1506 afterwards. This should put an end to stale PID files.
1508 * systemd-run will now also take relative binary path names
1509 for execution and no longer insists on absolute paths.
1511 * InaccessibleDirectories= and ReadOnlyDirectories= now take
1512 paths that are optionally prefixed with "-" to indicate that
1513 it should not be considered a failure if they do not exist.
1515 * journalctl -o (and similar commands) now understands a new
1516 output mode "short-precise", it is similar to "short" but
1517 shows timestamps with usec accuracy.
1519 * The option "discard" (as known from Debian) is now
1520 synonymous to "allow-discards" in /etc/crypttab. In fact,
1521 "discard" is preferred now (since it is easier to remember
1524 * Some licensing clean-ups were made, so that more code is now
1525 LGPL-2.1 licensed than before.
1527 * A minimal tool to save/restore the display backlight
1528 brightness across reboots has been added. It will store the
1529 backlight setting as late as possible at shutdown, and
1530 restore it as early as possible during reboot.
1532 * A logic to automatically discover and enable home and swap
1533 partitions on GPT disks has been added. With this in place
1534 /etc/fstab becomes optional for many setups as systemd can
1535 discover certain partitions located on the root disk
1536 automatically. Home partitions are recognized under their
1537 GPT type ID 933ac7e12eb44f13b8440e14e2aef915. Swap
1538 partitions are recognized under their GPT type ID
1539 0657fd6da4ab43c484e50933c84b4f4f.
1541 * systemd will no longer pass any environment from the kernel
1542 or initrd to system services. If you want to set an
1543 environment for all services, do so via the kernel command
1544 line systemd.setenv= assignment.
1546 * The systemd-sysctl tool no longer natively reads the file
1547 /etc/sysctl.conf. If desired, the file should be symlinked
1548 from /etc/sysctl.d/99-sysctl.conf. Apart from providing
1549 legacy support by a symlink rather than built-in code, it
1550 also makes the otherwise hidden order of application of the
1551 different files visible. (Note that this partly reverts to a
1552 pre-198 application order of sysctl knobs!)
1554 * The "systemctl set-log-level" and "systemctl dump" commands
1555 have been moved to systemd-analyze.
1557 * systemd-run learned the new --remain-after-exit switch,
1558 which causes the scope unit not to be cleaned up
1559 automatically after the process terminated.
1561 * tmpfiles learned a new --exclude-prefix= switch to exclude
1562 certain paths from operation.
1564 * journald will now automatically flush all messages to disk
1565 as soon as a message of the log priorities CRIT, ALERT or
1568 Contributions from: Andrew Cook, Brandon Philips, Christian
1569 Hesse, Christoph Junghans, Colin Walters, Daniel Schaal,
1570 Daniel Wallace, Dave Reisner, David Herrmann, Gao feng, George
1571 McCollister, Giovanni Campagna, Hannes Reinecke, Harald Hoyer,
1572 Herczeg Zsolt, Holger Hans Peter Freyther, Jan Engelhardt,
1573 Jesper Larsen, Kay Sievers, Khem Raj, Lennart Poettering,
1574 Lukas Nykryn, Maciej Wereski, Mantas Mikulėnas, Marcel
1575 Holtmann, Martin Pitt, Michael Biebl, Michael Marineau,
1576 Michael Scherer, Michael Stapelberg, Michal Sekletar, Michał
1577 Górny, Olivier Brunel, Ondrej Balaz, Ronny Chevalier, Shawn
1578 Landden, Steven Hiscocks, Thomas Bächler, Thomas Hindoe
1579 Paaboel Andersen, Tom Gundersen, Umut Tezduyar, WANG Chao,
1580 William Giokas, Zbigniew Jędrzejewski-Szmek
1582 -- Berlin, 2013-09-13
1586 * The documentation has been updated to cover the various new
1587 concepts introduced with 205.
1589 * Unit files now understand the new %v specifier which
1590 resolves to the kernel version string as returned by "uname
1593 * systemctl now supports filtering the unit list output by
1594 load state, active state and sub state, using the new
1597 * "systemctl status" will now show the results of the
1598 condition checks (like ConditionPathExists= and similar) of
1599 the last start attempts of the unit. They are also logged to
1602 * "journalctl -b" may now be used to look for boot output of a
1603 specific boot. Try "journalctl -b -1" for the previous boot,
1604 but the syntax is substantially more powerful.
1606 * "journalctl --show-cursor" has been added which prints the
1607 cursor string the last shown log line. This may then be used
1608 with the new "journalctl --after-cursor=" switch to continue
1609 browsing logs from that point on.
1611 * "journalctl --force" may now be used to force regeneration
1614 * Creation of "dead" device nodes has been moved from udev
1615 into kmod and tmpfiles. Previously, udev would read the kmod
1616 databases to pre-generate dead device nodes based on meta
1617 information contained in kernel modules, so that these would
1618 be auto-loaded on access rather then at boot. As this
1619 does not really have much to do with the exposing actual
1620 kernel devices to userspace this has always been slightly
1621 alien in the udev codebase. Following the new scheme kmod
1622 will now generate a runtime snippet for tmpfiles from the
1623 module meta information and it now is tmpfiles' job to the
1624 create the nodes. This also allows overriding access and
1625 other parameters for the nodes using the usual tmpfiles
1626 facilities. As side effect this allows us to remove the
1627 CAP_SYS_MKNOD capability bit from udevd entirely.
1629 * logind's device ACLs may now be applied to these "dead"
1630 devices nodes too, thus finally allowing managed access to
1631 devices such as /dev/snd/sequencer whithout loading the
1632 backing module right-away.
1634 * A new RPM macro has been added that may be used to apply
1635 tmpfiles configuration during package installation.
1637 * systemd-detect-virt and ConditionVirtualization= now can
1638 detect User-Mode-Linux machines (UML).
1640 * journald will now implicitly log the effective capabilities
1641 set of processes in the message metadata.
1643 * systemd-cryptsetup has gained support for TrueCrypt volumes.
1645 * The initrd interface has been simplified (more specifically,
1646 support for passing performance data via environment
1647 variables and fsck results via files in /run has been
1648 removed). These features were non-essential, and are
1649 nowadays available in a much nicer way by having systemd in
1650 the initrd serialize its state and have the hosts systemd
1651 deserialize it again.
1653 * The udev "keymap" data files and tools to apply keyboard
1654 specific mappings of scan to key codes, and force-release
1655 scan code lists have been entirely replaced by a udev
1656 "keyboard" builtin and a hwdb data file.
1658 * systemd will now honour the kernel's "quiet" command line
1659 argument also during late shutdown, resulting in a
1660 completely silent shutdown when used.
1662 * There's now an option to control the SO_REUSEPORT socket
1663 option in .socket units.
1665 * Instance units will now automatically get a per-template
1666 subslice of system.slice unless something else is explicitly
1667 configured. For example, instances of sshd@.service will now
1668 implicitly be placed in system-sshd.slice rather than
1669 system.slice as before.
1671 * Test coverage support may now be enabled at build time.
1673 Contributions from: Dave Reisner, Frederic Crozat, Harald
1674 Hoyer, Holger Hans Peter Freyther, Jan Engelhardt, Jan
1675 Janssen, Jason St. John, Jesper Larsen, Kay Sievers, Lennart
1676 Poettering, Lukas Nykryn, Maciej Wereski, Martin Pitt, Michael
1677 Olbrich, Ramkumar Ramachandra, Ross Lagerwall, Shawn Landden,
1678 Thomas H.P. Andersen, Tom Gundersen, Tomasz Torcz, William
1679 Giokas, Zbigniew Jędrzejewski-Szmek
1681 -- Berlin, 2013-07-23
1685 * Two new unit types have been introduced:
1687 Scope units are very similar to service units, however, are
1688 created out of pre-existing processes -- instead of PID 1
1689 forking off the processes. By using scope units it is
1690 possible for system services and applications to group their
1691 own child processes (worker processes) in a powerful way
1692 which then maybe used to organize them, or kill them
1693 together, or apply resource limits on them.
1695 Slice units may be used to partition system resources in an
1696 hierarchial fashion and then assign other units to them. By
1697 default there are now three slices: system.slice (for all
1698 system services), user.slice (for all user sessions),
1699 machine.slice (for VMs and containers).
1701 Slices and scopes have been introduced primarily in
1702 context of the work to move cgroup handling to a
1703 single-writer scheme, where only PID 1
1704 creates/removes/manages cgroups.
1706 * There's a new concept of "transient" units. In contrast to
1707 normal units these units are created via an API at runtime,
1708 not from configuration from disk. More specifically this
1709 means it is now possible to run arbitrary programs as
1710 independent services, with all execution parameters passed
1711 in via bus APIs rather than read from disk. Transient units
1712 make systemd substantially more dynamic then it ever was,
1713 and useful as a general batch manager.
1715 * logind has been updated to make use of scope and slice units
1716 for managing user sessions. As a user logs in he will get
1717 his own private slice unit, to which all sessions are added
1718 as scope units. We also added support for automatically
1719 adding an instance of user@.service for the user into the
1720 slice. Effectively logind will no longer create cgroup
1721 hierarchies on its own now, it will defer entirely to PID 1
1722 for this by means of scope, service and slice units. Since
1723 user sessions this way become entities managed by PID 1
1724 the output of "systemctl" is now a lot more comprehensive.
1726 * A new mini-daemon "systemd-machined" has been added which
1727 may be used by virtualization managers to register local
1728 VMs/containers. nspawn has been updated accordingly, and
1729 libvirt will be updated shortly. machined will collect a bit
1730 of meta information about the VMs/containers, and assign
1731 them their own scope unit (see above). The collected
1732 meta-data is then made available via the "machinectl" tool,
1733 and exposed in "ps" and similar tools. machined/machinectl
1734 is compile-time optional.
1736 * As discussed earlier, the low-level cgroup configuration
1737 options ControlGroup=, ControlGroupModify=,
1738 ControlGroupPersistent=, ControlGroupAttribute= have been
1739 removed. Please use high-level attribute settings instead as
1740 well as slice units.
1742 * A new bus call SetUnitProperties() has been added to alter
1743 various runtime parameters of a unit. This is primarily
1744 useful to alter cgroup parameters dynamically in a nice way,
1745 but will be extended later on to make more properties
1746 modifiable at runtime. systemctl gained a new set-properties
1747 command that wraps this call.
1749 * A new tool "systemd-run" has been added which can be used to
1750 run arbitrary command lines as transient services or scopes,
1751 while configuring a number of settings via the command
1752 line. This tool is currently very basic, however already
1753 very useful. We plan to extend this tool to even allow
1754 queuing of execution jobs with time triggers from the
1755 command line, similar in fashion to "at".
1757 * nspawn will now inform the user explicitly that kernels with
1758 audit enabled break containers, and suggest the user to turn
1761 * Support for detecting the IMA and AppArmor security
1762 frameworks with ConditionSecurity= has been added.
1764 * journalctl gained a new "-k" switch for showing only kernel
1765 messages, mimicking dmesg output; in addition to "--user"
1766 and "--system" switches for showing only user's own logs
1769 * systemd-delta can now show information about drop-in
1770 snippets extending unit files.
1772 * libsystemd-bus has been substantially updated but is still
1773 not available as public API.
1775 * systemd will now look for the "debug" argument on the kernel
1776 command line and enable debug logging, similar to what
1777 "systemd.log_level=debug" already did before.
1779 * "systemctl set-default", "systemctl get-default" has been
1780 added to configure the default.target symlink, which
1781 controls what to boot into by default.
1783 * "systemctl set-log-level" has been added as a convenient
1784 way to raise and lower systemd logging threshold.
1786 * "systemd-analyze plot" will now show the time the various
1787 generators needed for execution, as well as information
1788 about the unit file loading.
1790 * libsystemd-journal gained a new sd_journal_open_files() call
1791 for opening specific journal files. journactl also gained a
1792 new switch to expose this new functionality. Previously we
1793 only supported opening all files from a directory, or all
1794 files from the system, as opening individual files only is
1795 racy due to journal file rotation.
1797 * systemd gained the new DefaultEnvironment= setting in
1798 /etc/systemd/system.conf to set environment variables for
1801 * If a privileged process logs a journal message with the
1802 OBJECT_PID= field set, then journald will automatically
1803 augment this with additional OBJECT_UID=, OBJECT_GID=,
1804 OBJECT_COMM=, OBJECT_EXE=, ... fields. This is useful if
1805 system services want to log events about specific client
1806 processes. journactl/systemctl has been updated to make use
1807 of this information if all log messages regarding a specific
1810 Contributions from: Auke Kok, Chengwei Yang, Colin Walters,
1811 Cristian Rodríguez, Daniel Albers, Daniel Wallace, Dave
1812 Reisner, David Coppa, David King, David Strauss, Eelco
1813 Dolstra, Gabriel de Perthuis, Harald Hoyer, Jan Alexander
1814 Steffens, Jan Engelhardt, Jan Janssen, Jason St. John, Johan
1815 Heikkilä, Karel Zak, Karol Lewandowski, Kay Sievers, Lennart
1816 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marius Vollmer,
1817 Martin Pitt, Michael Biebl, Michael Olbrich, Michael Tremer,
1818 Michal Schmidt, Michał Bartoszkiewicz, Nirbheek Chauhan,
1819 Pierre Neidhardt, Ross Burton, Ross Lagerwall, Sean McGovern,
1820 Thomas Hindoe Paaboel Andersen, Tom Gundersen, Umut Tezduyar,
1821 Václav Pavlín, Zachary Cook, Zbigniew Jędrzejewski-Szmek,
1822 Łukasz Stelmach, 장동준
1826 * The Python bindings gained some minimal support for the APIs
1827 exposed by libsystemd-logind.
1829 * ConditionSecurity= gained support for detecting SMACK. Since
1830 this condition already supports SELinux and AppArmor we only
1831 miss IMA for this. Patches welcome!
1833 Contributions from: Karol Lewandowski, Lennart Poettering,
1834 Zbigniew Jędrzejewski-Szmek
1838 * systemd-nspawn will now create /etc/resolv.conf if
1839 necessary, before bind-mounting the host's file onto it.
1841 * systemd-nspawn will now store meta information about a
1842 container on the container's cgroup as extended attribute
1843 fields, including the root directory.
1845 * The cgroup hierarchy has been reworked in many ways. All
1846 objects any of the components systemd creates in the cgroup
1847 tree are now suffixed. More specifically, user sessions are
1848 now placed in cgroups suffixed with ".session", users in
1849 cgroups suffixed with ".user", and nspawn containers in
1850 cgroups suffixed with ".nspawn". Furthermore, all cgroup
1851 names are now escaped in a simple scheme to avoid collision
1852 of userspace object names with kernel filenames. This work
1853 is preparation for making these objects relocatable in the
1854 cgroup tree, in order to allow easy resource partitioning of
1855 these objects without causing naming conflicts.
1857 * systemctl list-dependencies gained the new switches
1858 --plain, --reverse, --after and --before.
1860 * systemd-inhibit now shows the process name of processes that
1861 have taken an inhibitor lock.
1863 * nss-myhostname will now also resolve "localhost"
1864 implicitly. This makes /etc/hosts an optional file and
1865 nicely handles that on IPv6 ::1 maps to both "localhost" and
1868 * libsystemd-logind.so gained a new call
1869 sd_get_machine_names() to enumerate running containers and
1870 VMs (currently only supported by very new libvirt and
1871 nspawn). sd_login_monitor can now be used to watch
1872 VMs/containers coming and going.
1874 * .include is not allowed recursively anymore, and only in
1875 unit files. Usually it is better to use drop-in snippets in
1876 .d/*.conf anyway, as introduced with systemd 198.
1878 * systemd-analyze gained a new "critical-chain" command that
1879 determines the slowest chain of units run during system
1880 boot-up. It is very useful for tracking down where
1881 optimizing boot time is the most beneficial.
1883 * systemd will no longer allow manipulating service paths in
1884 the name=systemd:/system cgroup tree using ControlGroup= in
1885 units. (But is still fine with it in all other dirs.)
1887 * There's a new systemd-nspawn@.service service file that may
1888 be used to easily run nspawn containers as system
1889 services. With the container's root directory in
1890 /var/lib/container/foobar it is now sufficient to run
1891 "systemctl start systemd-nspawn@foobar.service" to boot it.
1893 * systemd-cgls gained a new parameter "--machine" to list only
1894 the processes within a certain container.
1896 * ConditionSecurity= now can check for "apparmor". We still
1897 are lacking checks for SMACK and IMA for this condition
1898 check though. Patches welcome!
1900 * A new configuration file /etc/systemd/sleep.conf has been
1901 added that may be used to configure which kernel operation
1902 systemd is supposed to execute when "suspend", "hibernate"
1903 or "hybrid-sleep" is requested. This makes the new kernel
1904 "freeze" state accessible to the user.
1906 * ENV{SYSTEMD_WANTS} in udev rules will now implicitly escape
1907 the passed argument if applicable.
1909 Contributions from: Auke Kok, Colin Guthrie, Colin Walters,
1910 Cristian Rodríguez, Daniel Buch, Daniel Wallace, Dave Reisner,
1911 Evangelos Foutras, Greg Kroah-Hartman, Harald Hoyer, Josh
1912 Triplett, Kay Sievers, Lennart Poettering, Lukas Nykryn,
1913 MUNEDA Takahiro, Mantas Mikulėnas, Mirco Tischler, Nathaniel
1914 Chen, Nirbheek Chauhan, Ronny Chevalier, Ross Lagerwall, Tom
1915 Gundersen, Umut Tezduyar, Ville Skyttä, Zbigniew
1920 * The output of 'systemctl list-jobs' got some polishing. The
1921 '--type=' argument may now be passed more than once. A new
1922 command 'systemctl list-sockets' has been added which shows
1923 a list of kernel sockets systemd is listening on with the
1924 socket units they belong to, plus the units these socket
1927 * The experimental libsystemd-bus library got substantial
1928 updates to work in conjunction with the (also experimental)
1929 kdbus kernel project. It works well enough to exchange
1930 messages with some sophistication. Note that kdbus is not
1931 ready yet, and the library is mostly an elaborate test case
1932 for now, and not installable.
1934 * systemd gained a new unit 'systemd-static-nodes.service'
1935 that generates static device nodes earlier during boot, and
1936 can run in conjunction with udev.
1938 * libsystemd-login gained a new call sd_pid_get_user_unit()
1939 to retrieve the user systemd unit a process is running
1940 in. This is useful for systems where systemd is used as
1943 * systemd-nspawn now places all containers in the new /machine
1944 top-level cgroup directory in the name=systemd
1945 hierarchy. libvirt will soon do the same, so that we get a
1946 uniform separation of /system, /user and /machine for system
1947 services, user processes and containers/virtual
1948 machines. This new cgroup hierarchy is also useful to stick
1949 stable names to specific container instances, which can be
1950 recognized later this way (this name may be controlled
1951 via systemd-nspawn's new -M switch). libsystemd-login also
1952 gained a new call sd_pid_get_machine_name() to retrieve the
1953 name of the container/VM a specific process belongs to.
1955 * bootchart can now store its data in the journal.
1957 * libsystemd-journal gained a new call
1958 sd_journal_add_conjunction() for AND expressions to the
1959 matching logic. This can be used to express more complex
1960 logical expressions.
1962 * journactl can now take multiple --unit= and --user-unit=
1965 * The cryptsetup logic now understands the "luks.key=" kernel
1966 command line switch for specifying a file to read the
1967 decryption key from. Also, if a configured key file is not
1968 found the tool will now automatically fall back to prompting
1971 * Python systemd.journal module was updated to wrap recently
1972 added functions from libsystemd-journal. The interface was
1973 changed to bring the low level interface in s.j._Reader
1974 closer to the C API, and the high level interface in
1975 s.j.Reader was updated to wrap and convert all data about
1978 Contributions from: Anatol Pomozov, Auke Kok, Harald Hoyer,
1979 Henrik Grindal Bakken, Josh Triplett, Kay Sievers, Lennart
1980 Poettering, Lukas Nykryn, Mantas Mikulėnas Marius Vollmer,
1981 Martin Jansa, Martin Pitt, Michael Biebl, Michal Schmidt,
1982 Mirco Tischler, Pali Rohar, Simon Peeters, Steven Hiscocks,
1983 Tom Gundersen, Zbigniew Jędrzejewski-Szmek
1987 * journalctl --update-catalog now understands a new --root=
1988 option to operate on catalogs found in a different root
1991 * During shutdown after systemd has terminated all running
1992 services a final killing loop kills all remaining left-over
1993 processes. We will now print the name of these processes
1994 when we send SIGKILL to them, since this usually indicates a
1997 * If /etc/crypttab refers to password files stored on
1998 configured mount points automatic dependencies will now be
1999 generated to ensure the specific mount is established first
2000 before the key file is attempted to be read.
2002 * 'systemctl status' will now show information about the
2003 network sockets a socket unit is listening on.
2005 * 'systemctl status' will also shown information about any
2006 drop-in configuration file for units. (Drop-In configuration
2007 files in this context are files such as
2008 /etc/systemd/systemd/foobar.service.d/*.conf)
2010 * systemd-cgtop now optionally shows summed up CPU times of
2011 cgroups. Press '%' while running cgtop to switch between
2012 percentage and absolute mode. This is useful to determine
2013 which cgroups use up the most CPU time over the entire
2014 runtime of the system. systemd-cgtop has also been updated
2015 to be 'pipeable' for processing with further shell tools.
2017 * 'hostnamectl set-hostname' will now allow setting of FQDN
2020 * The formatting and parsing of time span values has been
2021 changed. The parser now understands fractional expressions
2022 such as "5.5h". The formatter will now output fractional
2023 expressions for all time spans under 1min, i.e. "5.123456s"
2024 rather than "5s 123ms 456us". For time spans under 1s
2025 millisecond values are shown, for those under 1ms
2026 microsecond values are shown. This should greatly improve
2027 all time-related output of systemd.
2029 * libsystemd-login and libsystemd-journal gained new
2030 functions for querying the poll() events mask and poll()
2031 timeout value for integration into arbitrary event
2034 * localectl gained the ability to list available X11 keymaps
2035 (models, layouts, variants, options).
2037 * 'systemd-analyze dot' gained the ability to filter for
2038 specific units via shell-style globs, to create smaller,
2039 more useful graphs. I.e. it is now possible to create simple
2040 graphs of all the dependencies between only target units, or
2041 of all units that Avahi has dependencies with.
2043 Contributions from: Cristian Rodríguez, Dr. Tilmann Bubeck,
2044 Harald Hoyer, Holger Hans Peter Freyther, Kay Sievers, Kelly
2045 Anderson, Koen Kooi, Lennart Poettering, Maksim Melnikau,
2046 Marc-Antoine Perennou, Marius Vollmer, Martin Pitt, Michal
2047 Schmidt, Oleksii Shevchuk, Ronny Chevalier, Simon McVittie,
2048 Steven Hiscocks, Thomas Weißschuh, Umut Tezduyar, Václav
2049 Pavlín, Zbigniew Jędrzejewski-Szmek, Łukasz Stelmach
2053 * The boot-time readahead implementation for rotating media
2054 will now read the read-ahead data in multiple passes which
2055 consist of all read requests made in equidistant time
2056 intervals. This means instead of strictly reading read-ahead
2057 data in its physical order on disk we now try to find a
2058 middle ground between physical and access time order.
2060 * /etc/os-release files gained a new BUILD_ID= field for usage
2061 on operating systems that provide continuous builds of OS
2064 Contributions from: Auke Kok, Eelco Dolstra, Kay Sievers,
2065 Lennart Poettering, Lukas Nykryn, Martin Pitt, Václav Pavlín
2066 William Douglas, Zbigniew Jędrzejewski-Szmek
2070 * systemd-python gained an API exposing libsystemd-daemon.
2072 * The SMACK setup logic gained support for uploading CIPSO
2075 * Behaviour of PrivateTmp=, ReadWriteDirectories=,
2076 ReadOnlyDirectories= and InaccessibleDirectories= has
2077 changed. The private /tmp and /var/tmp directories are now
2078 shared by all processes of a service (which means
2079 ExecStartPre= may now leave data in /tmp that ExecStart= of
2080 the same service can still access). When a service is
2081 stopped its temporary directories are immediately deleted
2082 (normal clean-up with tmpfiles is still done in addition to
2085 * By default, systemd will now set a couple of sysctl
2086 variables in the kernel: the safe sysrq options are turned
2087 on, IP route verification is turned on, and source routing
2088 disabled. The recently added hardlink and softlink
2089 protection of the kernel is turned on. These settings should
2090 be reasonably safe, and good defaults for all new systems.
2092 * The predictable network naming logic may now be turned off
2093 with a new kernel command line switch: net.ifnames=0.
2095 * A new libsystemd-bus module has been added that implements a
2096 pretty complete D-Bus client library. For details see:
2098 http://lists.freedesktop.org/archives/systemd-devel/2013-March/009797.html
2100 * journald will now explicitly flush the journal files to disk
2101 at the latest 5min after each write. The file will then also
2102 be marked offline until the next write. This should increase
2103 reliability in case of a crash. The synchronization delay
2104 can be configured via SyncIntervalSec= in journald.conf.
2106 * There's a new remote-fs-setup.target unit that can be used
2107 to pull in specific services when at least one remote file
2108 system is to be mounted.
2110 * There are new targets timers.target and paths.target as
2111 canonical targets to pull user timer and path units in
2112 from. This complements sockets.target with a similar
2113 purpose for socket units.
2115 * libudev gained a new call udev_device_set_attribute_value()
2116 to set sysfs attributes of a device.
2118 * The udev daemon now sets the default number of worker
2119 processes executed in parallel based on the number of available
2120 CPUs instead of the amount of available RAM. This is supposed
2121 to provide a more reliable default and limit a too aggressive
2122 paralellism for setups with 1000s of devices connected.
2124 Contributions from: Auke Kok, Colin Walters, Cristian
2125 Rodríguez, Daniel Buch, Dave Reisner, Frederic Crozat, Hannes
2126 Reinecke, Harald Hoyer, Jan Alexander Steffens, Jan
2127 Engelhardt, Josh Triplett, Kay Sievers, Lennart Poettering,
2128 Mantas Mikulėnas, Martin Pitt, Mathieu Bridon, Michael Biebl,
2129 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nathaniel Chen,
2130 Oleksii Shevchuk, Ozan Çağlayan, Thomas Hindoe Paaboel
2131 Andersen, Tollef Fog Heen, Tom Gundersen, Umut Tezduyar,
2132 Zbigniew Jędrzejewski-Szmek
2136 * Configuration of unit files may now be extended via drop-in
2137 files without having to edit/override the unit files
2138 themselves. More specifically, if the administrator wants to
2139 change one value for a service file foobar.service he can
2140 now do so by dropping in a configuration snippet into
2141 /etc/systemd/system/foobar.service.d/*.conf. The unit logic
2142 will load all these snippets and apply them on top of the
2143 main unit configuration file, possibly extending or
2144 overriding its settings. Using these drop-in snippets is
2145 generally nicer than the two earlier options for changing
2146 unit files locally: copying the files from
2147 /usr/lib/systemd/system/ to /etc/systemd/system/ and editing
2148 them there; or creating a new file in /etc/systemd/system/
2149 that incorporates the original one via ".include". Drop-in
2150 snippets into these .d/ directories can be placed in any
2151 directory systemd looks for units in, and the usual
2152 overriding semantics between /usr/lib, /etc and /run apply
2155 * Most unit file settings which take lists of items can now be
2156 reset by assigning the empty string to them. For example,
2157 normally, settings such as Environment=FOO=BAR append a new
2158 environment variable assignment to the environment block,
2159 each time they are used. By assigning Environment= the empty
2160 string the environment block can be reset to empty. This is
2161 particularly useful with the .d/*.conf drop-in snippets
2162 mentioned above, since this adds the ability to reset list
2163 settings from vendor unit files via these drop-ins.
2165 * systemctl gained a new "list-dependencies" command for
2166 listing the dependencies of a unit recursively.
2168 * Inhibitors are now honored and listed by "systemctl
2169 suspend", "systemctl poweroff" (and similar) too, not only
2170 GNOME. These commands will also list active sessions by
2173 * Resource limits (as exposed by the various control group
2174 controllers) can now be controlled dynamically at runtime
2175 for all units. More specifically, you can now use a command
2176 like "systemctl set-cgroup-attr foobar.service cpu.shares
2177 2000" to alter the CPU shares a specific service gets. These
2178 settings are stored persistently on disk, and thus allow the
2179 administrator to easily adjust the resource usage of
2180 services with a few simple commands. This dynamic resource
2181 management logic is also available to other programs via the
2182 bus. Almost any kernel cgroup attribute and controller is
2185 * systemd-vconsole-setup will now copy all font settings to
2186 all allocated VTs, where it previously applied them only to
2189 * libsystemd-login gained the new sd_session_get_tty() API
2192 * This release drops support for a few legacy or
2193 distribution-specific LSB facility names when parsing init
2194 scripts: $x-display-manager, $mail-transfer-agent,
2195 $mail-transport-agent, $mail-transfer-agent, $smtp,
2196 $null. Also, the mail-transfer-agent.target unit backing
2197 this has been removed. Distributions which want to retain
2198 compatibility with this should carry the burden for
2199 supporting this themselves and patch support for these back
2200 in, if they really need to. Also, the facilities $syslog and
2201 $local_fs are now ignored, since systemd does not support
2202 early-boot LSB init scripts anymore, and these facilities
2203 are implied anyway for normal services. syslog.target has
2206 * There are new bus calls on PID1's Manager object for
2207 cancelling jobs, and removing snapshot units. Previously,
2208 both calls were only available on the Job and Snapshot
2211 * systemd-journal-gatewayd gained SSL support.
2213 * The various "environment" files, such as /etc/locale.conf
2214 now support continuation lines with a backslash ("\") as
2215 last character in the line, similarly in style (but different)
2216 to how this is supported in shells.
2218 * For normal user processes the _SYSTEMD_USER_UNIT= field is
2219 now implicitly appended to every log entry logged. systemctl
2220 has been updated to filter by this field when operating on a
2221 user systemd instance.
2223 * nspawn will now implicitly add the CAP_AUDIT_WRITE and
2224 CAP_AUDIT_CONTROL capabilities to the capabilities set for
2225 the container. This makes it easier to boot unmodified
2226 Fedora systems in a container, which however still requires
2227 audit=0 to be passed on the kernel command line. Auditing in
2228 kernel and userspace is unfortunately still too broken in
2229 context of containers, hence we recommend compiling it out
2230 of the kernel or using audit=0. Hopefully this will be fixed
2231 one day for good in the kernel.
2233 * nspawn gained the new --bind= and --bind-ro= parameters to
2234 bind mount specific directories from the host into the
2237 * nspawn will now mount its own devpts file system instance
2238 into the container, in order not to leak pty devices from
2239 the host into the container.
2241 * systemd will now read the firmware boot time performance
2242 information from the EFI variables, if the used boot loader
2243 supports this, and takes it into account for boot performance
2244 analysis via "systemd-analyze". This is currently supported
2245 only in conjunction with Gummiboot, but could be supported
2246 by other boot loaders too. For details see:
2248 http://www.freedesktop.org/wiki/Software/systemd/BootLoaderInterface
2250 * A new generator has been added that automatically mounts the
2251 EFI System Partition (ESP) to /boot, if that directory
2252 exists, is empty, and no other file system has been
2253 configured to be mounted there.
2255 * logind will now send out PrepareForSleep(false) out
2256 unconditionally, after coming back from suspend. This may be
2257 used by applications as asynchronous notification for
2258 system resume events.
2260 * "systemctl unlock-sessions" has been added, that allows
2261 unlocking the screens of all user sessions at once, similar
2262 to how "systemctl lock-sessions" already locked all users
2263 sessions. This is backed by a new D-Bus call UnlockSessions().
2265 * "loginctl seat-status" will now show the master device of a
2266 seat. (i.e. the device of a seat that needs to be around for
2267 the seat to be considered available, usually the graphics
2270 * tmpfiles gained a new "X" line type, that allows
2271 configuration of files and directories (with wildcards) that
2272 shall be excluded from automatic cleanup ("aging").
2274 * udev default rules set the device node permissions now only
2275 at "add" events, and do not change them any longer with a
2276 later "change" event.
2278 * The log messages for lid events and power/sleep keypresses
2279 now carry a message ID.
2281 * We now have a substantially larger unit test suite, but this
2282 continues to be work in progress.
2284 * udevadm hwdb gained a new --root= parameter to change the
2285 root directory to operate relative to.
2287 * logind will now issue a background sync() request to the kernel
2288 early at shutdown, so that dirty buffers are flushed to disk early
2289 instead of at the last moment, in order to optimize shutdown
2292 * A new bootctl tool has been added that is an interface for
2293 certain boot loader operations. This is currently a preview
2294 and is likely to be extended into a small mechanism daemon
2295 like timedated, localed, hostnamed, and can be used by
2296 graphical UIs to enumerate available boot options, and
2297 request boot into firmware operations.
2299 * systemd-bootchart has been relicensed to LGPLv2.1+ to match
2300 the rest of the package. It also has been updated to work
2301 correctly in initrds.
2303 * Policykit previously has been runtime optional, and is now
2304 also compile time optional via a configure switch.
2306 * systemd-analyze has been reimplemented in C. Also "systemctl
2307 dot" has moved into systemd-analyze.
2309 * "systemctl status" with no further parameters will now print
2310 the status of all active or failed units.
2312 * Operations such as "systemctl start" can now be executed
2313 with a new mode "--irreversible" which may be used to queue
2314 operations that cannot accidentally be reversed by a later
2315 job queuing. This is by default used to make shutdown
2316 requests more robust.
2318 * The Python API of systemd now gained a new module for
2319 reading journal files.
2321 * A new tool kernel-install has been added that can install
2322 kernel images according to the Boot Loader Specification:
2324 http://www.freedesktop.org/wiki/Specifications/BootLoaderSpec
2326 * Boot time console output has been improved to provide
2327 animated boot time output for hanging jobs.
2329 * A new tool systemd-activate has been added which can be used
2330 to test socket activation with, directly from the command
2331 line. This should make it much easier to test and debug
2332 socket activation in daemons.
2334 * journalctl gained a new "--reverse" (or -r) option to show
2335 journal output in reverse order (i.e. newest line first).
2337 * journalctl gained a new "--pager-end" (or -e) option to jump
2338 to immediately jump to the end of the journal in the
2339 pager. This is only supported in conjunction with "less".
2341 * journalctl gained a new "--user-unit=" option, that works
2342 similarly to "--unit=" but filters for user units rather than
2345 * A number of unit files to ease adoption of systemd in
2346 initrds has been added. This moves some minimal logic from
2347 the various initrd implementations into systemd proper.
2349 * The journal files are now owned by a new group
2350 "systemd-journal", which exists specifically to allow access
2351 to the journal, and nothing else. Previously, we used the
2352 "adm" group for that, which however possibly covers more
2353 than just journal/log file access. This new group is now
2354 already used by systemd-journal-gatewayd to ensure this
2355 daemon gets access to the journal files and as little else
2356 as possible. Note that "make install" will also set FS ACLs
2357 up for /var/log/journal to give "adm" and "wheel" read
2358 access to it, in addition to "systemd-journal" which owns
2359 the journal files. We recommend that packaging scripts also
2360 add read access to "adm" + "wheel" to /var/log/journal, and
2361 all existing/future journal files. To normal users and
2362 administrators little changes, however packagers need to
2363 ensure to create the "systemd-journal" system group at
2364 package installation time.
2366 * The systemd-journal-gatewayd now runs as unprivileged user
2367 systemd-journal-gateway:systemd-journal-gateway. Packaging
2368 scripts need to create these system user/group at
2371 * timedated now exposes a new boolean property CanNTP that
2372 indicates whether a local NTP service is available or not.
2374 * systemd-detect-virt will now also detect xen PVs
2376 * The pstore file system is now mounted by default, if it is
2379 * In addition to the SELinux and IMA policies we will now also
2380 load SMACK policies at early boot.
2382 Contributions from: Adel Gadllah, Aleksander Morgado, Auke
2383 Kok, Ayan George, Bastien Nocera, Colin Walters, Daniel Buch,
2384 Daniel Wallace, Dave Reisner, David Herrmann, David Strauss,
2385 Eelco Dolstra, Enrico Scholz, Frederic Crozat, Harald Hoyer,
2386 Jan Janssen, Jonathan Callen, Kay Sievers, Lennart Poettering,
2387 Lukas Nykryn, Mantas Mikulėnas, Marc-Antoine Perennou, Martin
2388 Pitt, Mauro Dreissig, Max F. Albrecht, Michael Biebl, Michael
2389 Olbrich, Michal Schmidt, Michal Sekletar, Michal Vyskocil,
2390 Michał Bartoszkiewicz, Mirco Tischler, Nathaniel Chen, Nestor
2391 Ovroy, Oleksii Shevchuk, Paul W. Frields, Piotr Drąg, Rob
2392 Clark, Ryan Lortie, Simon McVittie, Simon Peeters, Steven
2393 Hiscocks, Thomas Hindoe Paaboel Andersen, Tollef Fog Heen, Tom
2394 Gundersen, Umut Tezduyar, William Giokas, Zbigniew
2395 Jędrzejewski-Szmek, Zeeshan Ali (Khattak)
2399 * Timer units now support calendar time events in addition to
2400 monotonic time events. That means you can now trigger a unit
2401 based on a calendar time specification such as "Thu,Fri
2402 2013-*-1,5 11:12:13" which refers to 11:12:13 of the first
2403 or fifth day of any month of the year 2013, given that it is
2404 a thursday or friday. This brings timer event support
2405 considerably closer to cron's capabilities. For details on
2406 the supported calendar time specification language see
2409 * udev now supports a number of different naming policies for
2410 network interfaces for predictable names, and a combination
2411 of these policies is now the default. Please see this wiki
2412 document for details:
2414 http://www.freedesktop.org/wiki/Software/systemd/PredictableNetworkInterfaceNames
2416 * Auke Kok's bootchart implementation has been added to the
2417 systemd tree. It is an optional component that can graph the
2418 boot in quite some detail. It is one of the best bootchart
2419 implementations around and minimal in its code and
2422 * nss-myhostname has been integrated into the systemd source
2423 tree. nss-myhostname guarantees that the local hostname
2424 always stays resolvable via NSS. It has been a weak
2425 requirement of systemd-hostnamed since a long time, and
2426 since its code is actually trivial we decided to just
2427 include it in systemd's source tree. It can be turned off
2428 with a configure switch.
2430 * The read-ahead logic is now capable of properly detecting
2431 whether a btrfs file system is on SSD or rotating media, in
2432 order to optimize the read-ahead scheme. Previously, it was
2433 only capable of detecting this on traditional file systems
2436 * In udev, additional device properties are now read from the
2437 IAB in addition to the OUI database. Also, Bluetooth company
2438 identities are attached to the devices as well.
2440 * In service files %U may be used as specifier that is
2441 replaced by the configured user name of the service.
2443 * nspawn may now be invoked without a controlling TTY. This
2444 makes it suitable for invocation as its own service. This
2445 may be used to set up a simple containerized server system
2446 using only core OS tools.
2448 * systemd and nspawn can now accept socket file descriptors
2449 when they are started for socket activation. This enables
2450 implementation of socket activated nspawn
2451 containers. i.e. think about autospawning an entire OS image
2452 when the first SSH or HTTP connection is received. We expect
2453 that similar functionality will also be added to libvirt-lxc
2456 * journalctl will now suppress ANSI color codes when
2457 presenting log data.
2459 * systemctl will no longer show control group information for
2460 a unit if a the control group is empty anyway.
2462 * logind can now automatically suspend/hibernate/shutdown the
2465 * /etc/machine-info and hostnamed now also expose the chassis
2466 type of the system. This can be used to determine whether
2467 the local system is a laptop, desktop, handset or
2468 tablet. This information may either be configured by the
2469 user/vendor or is automatically determined from ACPI and DMI
2470 information if possible.
2472 * A number of PolicyKit actions are now bound together with
2473 "imply" rules. This should simplify creating UIs because
2474 many actions will now authenticate similar ones as well.
2476 * Unit files learnt a new condition ConditionACPower= which
2477 may be used to conditionalize a unit depending on whether an
2478 AC power source is connected or not, of whether the system
2479 is running on battery power.
2481 * systemctl gained a new "is-failed" verb that may be used in
2482 shell scripts and suchlike to check whether a specific unit
2483 is in the "failed" state.
2485 * The EnvironmentFile= setting in unit files now supports file
2486 globbing, and can hence be used to easily read a number of
2487 environment files at once.
2489 * systemd will no longer detect and recognize specific
2490 distributions. All distribution-specific #ifdeffery has been
2491 removed, systemd is now fully generic and
2492 distribution-agnostic. Effectively, not too much is lost as
2493 a lot of the code is still accessible via explicit configure
2494 switches. However, support for some distribution specific
2495 legacy configuration file formats has been dropped. We
2496 recommend distributions to simply adopt the configuration
2497 files everybody else uses now and convert the old
2498 configuration from packaging scripts. Most distributions
2499 already did that. If that's not possible or desirable,
2500 distributions are welcome to forward port the specific
2501 pieces of code locally from the git history.
2503 * When logging a message about a unit systemd will now always
2504 log the unit name in the message meta data.
2506 * localectl will now also discover system locale data that is
2507 not stored in locale archives, but directly unpacked.
2509 * logind will no longer unconditionally use framebuffer
2510 devices as seat masters, i.e. as devices that are required
2511 to be existing before a seat is considered preset. Instead,
2512 it will now look for all devices that are tagged as
2513 "seat-master" in udev. By default framebuffer devices will
2514 be marked as such, but depending on local systems other
2515 devices might be marked as well. This may be used to
2516 integrate graphics cards using closed source drivers (such
2517 as NVidia ones) more nicely into logind. Note however, that
2518 we recommend using the open source NVidia drivers instead,
2519 and no udev rules for the closed-source drivers will be
2520 shipped from us upstream.
2522 Contributions from: Adam Williamson, Alessandro Crismani, Auke
2523 Kok, Colin Walters, Daniel Wallace, Dave Reisner, David
2524 Herrmann, David Strauss, Dimitrios Apostolou, Eelco Dolstra,
2525 Eric Benoit, Giovanni Campagna, Hannes Reinecke, Henrik
2526 Grindal Bakken, Hermann Gausterer, Kay Sievers, Lennart
2527 Poettering, Lukas Nykryn, Mantas Mikulėnas, Marcel Holtmann,
2528 Martin Pitt, Matthew Monaco, Michael Biebl, Michael Terry,
2529 Michal Schmidt, Michal Sekletar, Michał Bartoszkiewicz, Oleg
2530 Samarin, Pekka Lundstrom, Philip Nilsson, Ramkumar
2531 Ramachandra, Richard Yao, Robert Millan, Sami Kerola, Shawn
2532 Landden, Thomas Hindoe Paaboel Andersen, Thomas Jarosch,
2533 Tollef Fog Heen, Tom Gundersen, Umut Tezduyar, Zbigniew
2538 * udev gained support for loading additional device properties
2539 from an indexed database that is keyed by vendor/product IDs
2540 and similar device identifiers. For the beginning this
2541 "hwdb" is populated with data from the well-known PCI and
2542 USB database, but also includes PNP, ACPI and OID data. In
2543 the longer run this indexed database shall grow into
2544 becoming the one central database for non-essential
2545 userspace device metadata. Previously, data from the PCI/USB
2546 database was only attached to select devices, since the
2547 lookup was a relatively expensive operation due to O(n) time
2548 complexity (with n being the number of entries in the
2549 database). Since this is now O(1), we decided to add in this
2550 data for all devices where this is available, by
2551 default. Note that the indexed database needs to be rebuilt
2552 when new data files are installed. To achieve this you need
2553 to update your packaging scripts to invoke "udevadm hwdb
2554 --update" after installation of hwdb data files. For
2555 RPM-based distributions we introduced the new
2556 %udev_hwdb_update macro for this purpose.
2558 * The Journal gained support for the "Message Catalog", an
2559 indexed database to link up additional information with
2560 journal entries. For further details please check:
2562 http://www.freedesktop.org/wiki/Software/systemd/catalog
2564 The indexed message catalog database also needs to be
2565 rebuilt after installation of message catalog files. Use
2566 "journalctl --update-catalog" for this. For RPM-based
2567 distributions we introduced the %journal_catalog_update
2568 macro for this purpose.
2570 * The Python Journal bindings gained support for the standard
2571 Python logging framework.
2573 * The Journal API gained new functions for checking whether
2574 the underlying file system of a journal file is capable of
2575 properly reporting file change notifications, or whether
2576 applications that want to reflect journal changes "live"
2577 need to recheck journal files continuously in appropriate
2580 * It is now possible to set the "age" field for tmpfiles
2581 entries to 0, indicating that files matching this entry
2582 shall always be removed when the directories are cleaned up.
2584 * coredumpctl gained a new "gdb" verb which invokes gdb
2585 right-away on the selected coredump.
2587 * There's now support for "hybrid sleep" on kernels that
2588 support this, in addition to "suspend" and "hibernate". Use
2589 "systemctl hybrid-sleep" to make use of this.
2591 * logind's HandleSuspendKey= setting (and related settings)
2592 now gained support for a new "lock" setting to simply
2593 request the screen lock on all local sessions, instead of
2594 actually executing a suspend or hibernation.
2596 * systemd will now mount the EFI variables file system by
2599 * Socket units now gained support for configuration of the
2600 SMACK security label.
2602 * timedatectl will now output the time of the last and next
2603 daylight saving change.
2605 * We dropped support for various legacy and distro-specific
2606 concepts, such as insserv, early-boot SysV services
2607 (i.e. those for non-standard runlevels such as 'b' or 'S')
2608 or ArchLinux /etc/rc.conf support. We recommend the
2609 distributions who still need support this to either continue
2610 to maintain the necessary patches downstream, or find a
2611 different solution. (Talk to us if you have questions!)
2613 * Various systemd components will now bypass PolicyKit checks
2614 for root and otherwise handle properly if PolicyKit is not
2615 found to be around. This should fix most issues for
2616 PolicyKit-less systems. Quite frankly this should have been
2617 this way since day one. It is absolutely our intention to
2618 make systemd work fine on PolicyKit-less systems, and we
2619 consider it a bug if something does not work as it should if
2620 PolicyKit is not around.
2622 * For embedded systems it is now possible to build udev and
2623 systemd without blkid and/or kmod support.
2625 * "systemctl switch-root" is now capable of switching root
2626 more than once. I.e. in addition to transitions from the
2627 initrd to the host OS it is now possible to transition to
2628 further OS images from the host. This is useful to implement
2629 offline updating tools.
2631 * Various other additions have been made to the RPM macros
2632 shipped with systemd. Use %udev_rules_update() after
2633 installing new udev rules files. %_udevhwdbdir,
2634 %_udevrulesdir, %_journalcatalogdir, %_tmpfilesdir,
2635 %_sysctldir are now available which resolve to the right
2636 directories for packages to place various data files in.
2638 * journalctl gained the new --full switch (in addition to
2639 --all, to disable ellipsation for long messages.
2641 Contributions from: Anders Olofsson, Auke Kok, Ben Boeckel,
2642 Colin Walters, Cosimo Cecchi, Daniel Wallace, Dave Reisner,
2643 Eelco Dolstra, Holger Hans Peter Freyther, Kay Sievers,
2644 Chun-Yi Lee, Lekensteyn, Lennart Poettering, Mantas Mikulėnas,
2645 Marti Raudsepp, Martin Pitt, Mauro Dreissig, Michael Biebl,
2646 Michal Schmidt, Michal Sekletar, Miklos Vajna, Nis Martensen,
2647 Oleksii Shevchuk, Olivier Brunel, Ramkumar Ramachandra, Thomas
2648 Bächler, Thomas Hindoe Paaboel Andersen, Tom Gundersen, Tony
2649 Camuso, Umut Tezduyar, Zbigniew Jędrzejewski-Szmek
2653 * journalctl gained new --since= and --until= switches to
2654 filter by time. It also now supports nice filtering for
2655 units via --unit=/-u.
2657 * Type=oneshot services may use ExecReload= and do the
2660 * The journal daemon now supports time-based rotation and
2661 vacuuming, in addition to the usual disk-space based
2664 * The journal will now index the available field values for
2665 each field name. This enables clients to show pretty drop
2666 downs of available match values when filtering. The bash
2667 completion of journalctl has been updated
2668 accordingly. journalctl gained a new switch -F to list all
2669 values a certain field takes in the journal database.
2671 * More service events are now written as structured messages
2672 to the journal, and made recognizable via message IDs.
2674 * The timedated, localed and hostnamed mini-services which
2675 previously only provided support for changing time, locale
2676 and hostname settings from graphical DEs such as GNOME now
2677 also have a minimal (but very useful) text-based client
2678 utility each. This is probably the nicest way to changing
2679 these settings from the command line now, especially since
2680 it lists available options and is fully integrated with bash
2683 * There's now a new tool "systemd-coredumpctl" to list and
2684 extract coredumps from the journal.
2686 * We now install a README each in /var/log/ and
2687 /etc/rc.d/init.d explaining where the system logs and init
2688 scripts went. This hopefully should help folks who go to
2689 that dirs and look into the otherwise now empty void and
2690 scratch their heads.
2692 * When user-services are invoked (by systemd --user) the
2693 $MANAGERPID env var is set to the PID of systemd.
2695 * SIGRTMIN+24 when sent to a --user instance will now result
2696 in immediate termination of systemd.
2698 * gatewayd received numerous feature additions such as a
2699 "follow" mode, for live syncing and filtering.
2701 * browse.html now allows filtering and showing detailed
2702 information on specific entries. Keyboard navigation and
2703 mouse screen support has been added.
2705 * gatewayd/journalctl now supports HTML5/JSON
2706 Server-Sent-Events as output.
2708 * The SysV init script compatibility logic will now
2709 heuristically determine whether a script supports the
2710 "reload" verb, and only then make this available as
2713 * "systemctl status --follow" has been removed, use "journalctl
2716 * journald.conf's RuntimeMinSize=, PersistentMinSize= settings
2717 have been removed since they are hardly useful to be
2720 * And I'd like to take the opportunity to specifically mention
2721 Zbigniew for his great contributions. Zbigniew, you rock!
2723 Contributions from: Andrew Eikum, Christian Hesse, Colin
2724 Guthrie, Daniel J Walsh, Dave Reisner, Eelco Dolstra, Ferenc
2725 Wágner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Mantas
2726 Mikulėnas, Martin Mikkelsen, Martin Pitt, Michael Olbrich,
2727 Michael Stapelberg, Michal Schmidt, Sebastian Ott, Thomas
2728 Bächler, Umut Tezduyar, Will Woods, Wulf C. Krueger, Zbigniew
2729 Jędrzejewski-Szmek, Сковорода Никита Андреевич
2733 * If /etc/vconsole.conf is non-existent or empty we will no
2734 longer load any console font or key map at boot by
2735 default. Instead the kernel defaults will be left
2736 intact. This is definitely the right thing to do, as no
2737 configuration should mean no configuration, and hard-coding
2738 font names that are different on all archs is probably a bad
2739 idea. Also, the kernel default key map and font should be
2740 good enough for most cases anyway, and mostly identical to
2741 the userspace fonts/key maps we previously overloaded them
2742 with. If distributions want to continue to default to a
2743 non-kernel font or key map they should ship a default
2744 /etc/vconsole.conf with the appropriate contents.
2746 Contributions from: Colin Walters, Daniel J Walsh, Dave
2747 Reisner, Kay Sievers, Lennart Poettering, Lukas Nykryn, Tollef
2748 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek
2752 * journalctl gained a new --cursor= switch to show entries
2753 starting from the specified location in the journal.
2755 * We now enforce a size limit on journal entry fields exported
2756 with "-o json" in journalctl. Fields larger than 4K will be
2757 assigned null. This can be turned off with --all.
2759 * An (optional) journal gateway daemon is now available as
2760 "systemd-journal-gatewayd.service". This service provides
2761 access to the journal via HTTP and JSON. This functionality
2762 will be used to implement live log synchronization in both
2763 pull and push modes, but has various other users too, such
2764 as easy log access for debugging of embedded devices. Right
2765 now it is already useful to retrieve the journal via HTTP:
2767 # systemctl start systemd-journal-gatewayd.service
2768 # wget http://localhost:19531/entries
2770 This will download the journal contents in a
2771 /var/log/messages compatible format. The same as JSON:
2773 # curl -H"Accept: application/json" http://localhost:19531/entries
2775 This service is also accessible via a web browser where a
2776 single static HTML5 app is served that uses the JSON logic
2777 to enable the user to do some basic browsing of the
2778 journal. This will be extended later on. Here's an example
2779 screenshot of this app in its current state:
2781 http://0pointer.de/public/journal-gatewayd
2783 Contributions from: Kay Sievers, Lennart Poettering, Robert
2784 Milasan, Tom Gundersen
2788 * The bash completion logic is now available for journalctl
2791 * We do not mount the "cpuset" controller anymore together with
2792 "cpu" and "cpuacct", as "cpuset" groups generally cannot be
2793 started if no parameters are assigned to it. "cpuset" hence
2794 broke code that assumed it it could create "cpu" groups and
2797 * journalctl -f will now subscribe to terminal size changes,
2798 and line break accordingly.
2800 Contributions from: Dave Reisner, Kay Sievers, Lennart
2801 Poettering, Lukas Nykrynm, Mirco Tischler, Václav Pavlín
2805 * nspawn will now create a symlink /etc/localtime in the
2806 container environment, copying the host's timezone
2807 setting. Previously this has been done via a bind mount, but
2808 since symlinks cannot be bind mounted this has now been
2809 changed to create/update the appropriate symlink.
2811 * journalctl -n's line number argument is now optional, and
2812 will default to 10 if omitted.
2814 * journald will now log the maximum size the journal files may
2815 take up on disk. This is particularly useful if the default
2816 built-in logic of determining this parameter from the file
2817 system size is used. Use "systemctl status
2818 systemd-journald.service" to see this information.
2820 * The multi-seat X wrapper tool has been stripped down. As X
2821 is now capable of enumerating graphics devices via udev in a
2822 seat-aware way the wrapper is not strictly necessary
2823 anymore. A stripped down temporary stop-gap is still shipped
2824 until the upstream display managers have been updated to
2825 fully support the new X logic. Expect this wrapper to be
2826 removed entirely in one of the next releases.
2828 * HandleSleepKey= in logind.conf has been split up into
2829 HandleSuspendKey= and HandleHibernateKey=. The old setting
2830 is not available anymore. X11 and the kernel are
2831 distuingishing between these keys and we should too. This
2832 also means the inhibition lock for these keys has been split
2835 Contributions from: Dave Airlie, Eelco Dolstra, Lennart
2836 Poettering, Lukas Nykryn, Václav Pavlín
2840 * Whenever a unit changes state we will now log this to the
2841 journal and show along the unit's own log output in
2844 * ConditionPathIsMountPoint= can now properly detect bind
2845 mount points too. (Previously, a bind mount of one file
2846 system to another place in the same file system could not be
2847 detected as mount, since they shared struct stat's st_dev
2850 * We will now mount the cgroup controllers cpu, cpuacct,
2851 cpuset and the controllers net_cls, net_prio together by
2854 * nspawn containers will now have a virtualized boot
2855 ID. (i.e. /proc/sys/kernel/random/boot_id is now mounted
2856 over with a randomized ID at container initialization). This
2857 has the effect of making "journalctl -b" do the right thing
2860 * The JSON output journal serialization has been updated not
2861 to generate "endless" list objects anymore, but rather one
2862 JSON object per line. This is more in line how most JSON
2863 parsers expect JSON objects. The new output mode
2864 "json-pretty" has been added to provide similar output, but
2865 neatly aligned for readability by humans.
2867 * We dropped all explicit sync() invocations in the shutdown
2868 code. The kernel does this implicitly anyway in the kernel
2869 reboot() syscall. halt(8)'s -n option is now a compatibility
2872 * We now support virtualized reboot() in containers, as
2873 supported by newer kernels. We will fall back to exit() if
2874 CAP_SYS_REBOOT is not available to the container. Also,
2875 nspawn makes use of this now and will actually reboot the
2876 container if the containerized OS asks for that.
2878 * journalctl will only show local log output by default
2879 now. Use --merge (-m) to show remote log output, too.
2881 * libsystemd-journal gained the new sd_journal_get_usage()
2882 call to determine the current disk usage of all journal
2883 files. This is exposed in the new "journalctl --disk-usage"
2886 * journald gained a new configuration setting SplitMode= in
2887 journald.conf which may be used to control how user journals
2888 are split off. See journald.conf(5) for details.
2890 * A new condition type ConditionFileNotEmpty= has been added.
2892 * tmpfiles' "w" lines now support file globbing, to write
2893 multiple files at once.
2895 * We added Python bindings for the journal submission
2896 APIs. More Python APIs for a number of selected APIs will
2897 likely follow. Note that we intend to add native bindings
2898 only for the Python language, as we consider it common
2899 enough to deserve bindings shipped within systemd. There are
2900 various projects outside of systemd that provide bindings
2901 for languages such as PHP or Lua.
2903 * Many conditions will now resolve specifiers such as %i. In
2904 addition, PathChanged= and related directives of .path units
2905 now support specifiers as well.
2907 * There's now a new RPM macro definition for the system preset
2910 * journald will now warn if it ca not forward a message to the
2911 syslog daemon because its socket is full.
2913 * timedated will no longer write or process /etc/timezone,
2914 except on Debian. As we do not support late mounted /usr
2915 anymore /etc/localtime always being a symlink is now safe,
2916 and hence the information in /etc/timezone is not necessary
2919 * logind will now always reserve one VT for a text getty (VT6
2920 by default). Previously if more than 6 X sessions where
2921 started they took up all the VTs with auto-spawned gettys,
2922 so that no text gettys were available anymore.
2924 * udev will now automatically inform the btrfs kernel logic
2925 about btrfs RAID components showing up. This should make
2926 simple hotplug based btrfs RAID assembly work.
2928 * PID 1 will now increase its RLIMIT_NOFILE to 64K by default
2929 (but not for its children which will stay at the kernel
2930 default). This should allow setups with a lot more listening
2933 * systemd will now always pass the configured timezone to the
2934 kernel at boot. timedated will do the same when the timezone
2937 * logind's inhibition logic has been updated. By default,
2938 logind will now handle the lid switch, the power and sleep
2939 keys all the time, even in graphical sessions. If DEs want
2940 to handle these events on their own they should take the new
2941 handle-power-key, handle-sleep-key and handle-lid-switch
2942 inhibitors during their runtime. A simple way to achiveve
2943 that is to invoke the DE wrapped in an invocation of:
2945 systemd-inhibit --what=handle-power-key:handle-sleep-key:handle-lid-switch ...
2947 * Access to unit operations is now checked via SELinux taking
2948 the unit file label and client process label into account.
2950 * systemd will now notify the administrator in the journal
2951 when he over-mounts a non-empty directory.
2953 * There are new specifiers that are resolved in unit files,
2954 for the host name (%H), the machine ID (%m) and the boot ID
2957 Contributions from: Allin Cottrell, Auke Kok, Brandon Philips,
2958 Colin Guthrie, Colin Walters, Daniel J Walsh, Dave Reisner,
2959 Eelco Dolstra, Jan Engelhardt, Kay Sievers, Lennart
2960 Poettering, Lucas De Marchi, Lukas Nykryn, Mantas Mikulėnas,
2961 Martin Pitt, Matthias Clasen, Michael Olbrich, Pierre Schmitz,
2962 Shawn Landden, Thomas Hindoe Paaboel Andersen, Tom Gundersen,
2963 Václav Pavlín, Yin Kangkai, Zbigniew Jędrzejewski-Szmek
2967 * Support for reading structured kernel messages from
2968 /dev/kmsg has now been added and is enabled by default.
2970 * Support for reading kernel messages from /proc/kmsg has now
2971 been removed. If you want kernel messages in the journal
2972 make sure to run a recent kernel (>= 3.5) that supports
2973 reading structured messages from /dev/kmsg (see
2974 above). /proc/kmsg is now exclusive property of classic
2975 syslog daemons again.
2977 * The libudev API gained the new
2978 udev_device_new_from_device_id() call.
2980 * The logic for file system namespace (ReadOnlyDirectory=,
2981 ReadWriteDirectoy=, PrivateTmp=) has been reworked not to
2982 require pivot_root() anymore. This means fewer temporary
2983 directories are created below /tmp for this feature.
2985 * nspawn containers will now see and receive all submounts
2986 made on the host OS below the root file system of the
2989 * Forward Secure Sealing is now supported for Journal files,
2990 which provide cryptographical sealing of journal files so
2991 that attackers cannot alter log history anymore without this
2992 being detectable. Lennart will soon post a blog story about
2993 this explaining it in more detail.
2995 * There are two new service settings RestartPreventExitStatus=
2996 and SuccessExitStatus= which allow configuration of exit
2997 status (exit code or signal) which will be excepted from the
2998 restart logic, resp. consider successful.
3000 * journalctl gained the new --verify switch that can be used
3001 to check the integrity of the structure of journal files and
3002 (if Forward Secure Sealing is enabled) the contents of
3005 * nspawn containers will now be run with /dev/stdin, /dev/fd/
3006 and similar symlinks pre-created. This makes running shells
3007 as container init process a lot more fun.
3009 * The fstab support can now handle PARTUUID= and PARTLABEL=
3012 * A new ConditionHost= condition has been added to match
3013 against the hostname (with globs) and machine ID. This is
3014 useful for clusters where a single OS image is used to
3015 provision a large number of hosts which shall run slightly
3016 different sets of services.
3018 * Services which hit the restart limit will now be placed in a
3021 Contributions from: Bertram Poettering, Dave Reisner, Huang
3022 Hang, Kay Sievers, Lennart Poettering, Lukas Nykryn, Martin
3023 Pitt, Simon Peeters, Zbigniew Jędrzejewski-Szmek
3027 * When running in --user mode systemd will now become a
3028 subreaper (PR_SET_CHILD_SUBREAPER). This should make the ps
3029 tree a lot more organized.
3031 * A new PartOf= unit dependency type has been introduced that
3032 may be used to group services in a natural way.
3034 * "systemctl enable" may now be used to enable instances of
3037 * journalctl now prints error log levels in red, and
3038 warning/notice log levels in bright white. It also supports
3039 filtering by log level now.
3041 * cgtop gained a new -n switch (similar to top), to configure
3042 the maximum number of iterations to run for. It also gained
3043 -b, to run in batch mode (accepting no input).
3045 * The suffix ".service" may now be omitted on most systemctl
3046 command lines involving service unit names.
3048 * There's a new bus call in logind to lock all sessions, as
3049 well as a loginctl verb for it "lock-sessions".
3051 * libsystemd-logind.so gained a new call sd_journal_perror()
3052 that works similar to libc perror() but logs to the journal
3053 and encodes structured information about the error number.
3055 * /etc/crypttab entries now understand the new keyfile-size=
3058 * shutdown(8) now can send a (configurable) wall message when
3059 a shutdown is cancelled.
3061 * The mount propagation mode for the root file system will now
3062 default to "shared", which is useful to make containers work
3063 nicely out-of-the-box so that they receive new mounts from
3064 the host. This can be undone locally by running "mount
3065 --make-rprivate /" if needed.
3067 * The prefdm.service file has been removed. Distributions
3068 should maintain this unit downstream if they intend to keep
3069 it around. However, we recommend writing normal unit files
3070 for display managers instead.
3072 * Since systemd is a crucial part of the OS we will now
3073 default to a number of compiler switches that improve
3074 security (hardening) such as read-only relocations, stack
3075 protection, and suchlike.
3077 * The TimeoutSec= setting for services is now split into
3078 TimeoutStartSec= and TimeoutStopSec= to allow configuration
3079 of individual time outs for the start and the stop phase of
3082 Contributions from: Artur Zaprzala, Arvydas Sidorenko, Auke
3083 Kok, Bryan Kadzban, Dave Reisner, David Strauss, Harald Hoyer,
3084 Jim Meyering, Kay Sievers, Lennart Poettering, Mantas
3085 Mikulėnas, Martin Pitt, Michal Schmidt, Michal Sekletar, Peter
3086 Alfredsen, Shawn Landden, Simon Peeters, Terence Honles, Tom
3087 Gundersen, Zbigniew Jędrzejewski-Szmek
3091 * The journal and id128 C APIs are now fully documented as man
3094 * Extra safety checks have been added when transitioning from
3095 the initial RAM disk to the main system to avoid accidental
3098 * /etc/crypttab entries now understand the new keyfile-offset=
3101 * systemctl -t can now be used to filter by unit load state.
3103 * The journal C API gained the new sd_journal_wait() call to
3104 make writing synchronous journal clients easier.
3106 * journalctl gained the new -D switch to show journals from a
3109 * journalctl now displays a special marker between log
3110 messages of two different boots.
3112 * The journal is now explicitly flushed to /var via a service
3113 systemd-journal-flush.service, rather than implicitly simply
3114 by seeing /var/log/journal to be writable.
3116 * journalctl (and the journal C APIs) can now match for much
3117 more complex expressions, with alternatives and
3120 * When transitioning from the initial RAM disk to the main
3121 system we will now kill all processes in a killing spree to
3122 ensure no processes stay around by accident.
3124 * Three new specifiers may be used in unit files: %u, %h, %s
3125 resolve to the user name, user home directory resp. user
3126 shell. This is useful for running systemd user instances.
3128 * We now automatically rotate journal files if their data
3129 object hash table gets a fill level > 75%. We also size the
3130 hash table based on the configured maximum file size. This
3131 together should lower hash collisions drastically and thus
3132 speed things up a bit.
3134 * journalctl gained the new "--header" switch to introspect
3135 header data of journal files.
3137 * A new setting SystemCallFilters= has been added to services
3138 which may be used to apply blacklists or whitelists to
3139 system calls. This is based on SECCOMP Mode 2 of Linux 3.5.
3141 * nspawn gained a new --link-journal= switch (and quicker: -j)
3142 to link the container journal with the host. This makes it
3143 very easy to centralize log viewing on the host for all
3144 guests while still keeping the journal files separated.
3146 * Many bugfixes and optimizations
3148 Contributions from: Auke Kok, Eelco Dolstra, Harald Hoyer, Kay
3149 Sievers, Lennart Poettering, Malte Starostik, Paul Menzel, Rex
3150 Tsai, Shawn Landden, Tom Gundersen, Ville Skyttä, Zbigniew
3155 * Several tools now understand kernel command line arguments,
3156 which are only read when run in an initial RAM disk. They
3157 usually follow closely their normal counterparts, but are
3160 * There's a new tool to analyze the readahead files that are
3161 automatically generated at boot. Use:
3163 /usr/lib/systemd/systemd-readahead analyze /.readahead
3165 * We now provide an early debug shell on tty9 if this enabled. Use:
3167 systemctl enable debug-shell.service
3169 * All plymouth related units have been moved into the Plymouth
3170 package. Please make sure to upgrade your Plymouth version
3173 * systemd-tmpfiles now supports getting passed the basename of
3174 a configuration file only, in which case it will look for it
3175 in all appropriate directories automatically.
3177 * udevadm info now takes a /dev or /sys path as argument, and
3178 does the right thing. Example:
3180 udevadm info /dev/sda
3181 udevadm info /sys/class/block/sda
3183 * systemctl now prints a warning if a unit is stopped but a
3184 unit that might trigger it continues to run. Example: a
3185 service is stopped but the socket that activates it is left
3188 * "systemctl status" will now mention if the log output was
3189 shortened due to rotation since a service has been started.
3191 * The journal API now exposes functions to determine the
3192 "cutoff" times due to rotation.
3194 * journald now understands SIGUSR1 and SIGUSR2 for triggering
3195 immediately flushing of runtime logs to /var if possible,
3196 resp. for triggering immediate rotation of the journal
3199 * It is now considered an error if a service is attempted to
3200 be stopped that is not loaded.
3202 * XDG_RUNTIME_DIR now uses numeric UIDs instead of usernames.
3204 * systemd-analyze now supports Python 3
3206 * tmpfiles now supports cleaning up directories via aging
3207 where the first level dirs are always kept around but
3208 directories beneath it automatically aged. This is enabled
3209 by prefixing the age field with '~'.
3211 * Seat objects now expose CanGraphical, CanTTY properties
3212 which is required to deal with very fast bootups where the
3213 display manager might be running before the graphics drivers
3214 completed initialization.
3216 * Seat objects now expose a State property.
3218 * We now include RPM macros for service enabling/disabling
3219 based on the preset logic. We recommend RPM based
3220 distributions to make use of these macros if possible. This
3221 makes it simpler to reuse RPM spec files across
3224 * We now make sure that the collected systemd unit name is
3225 always valid when services log to the journal via
3228 * There's a new man page kernel-command-line(7) detailing all
3229 command line options we understand.
3231 * The fstab generator may now be disabled at boot by passing
3232 fstab=0 on the kernel command line.
3234 * A new kernel command line option modules-load= is now understood
3235 to load a specific kernel module statically, early at boot.
3237 * Unit names specified on the systemctl command line are now
3238 automatically escaped as needed. Also, if file system or
3239 device paths are specified they are automatically turned
3240 into the appropriate mount or device unit names. Example:
3242 systemctl status /home
3243 systemctl status /dev/sda
3245 * The SysVConsole= configuration option has been removed from
3246 system.conf parsing.
3248 * The SysV search path is no longer exported on the D-Bus
3251 * The Names= option is been removed from unit file parsing.
3253 * There's a new man page bootup(7) detailing the boot process.
3255 * Every unit and every generator we ship with systemd now
3256 comes with full documentation. The self-explanatory boot is
3259 * A couple of services gained "systemd-" prefixes in their
3260 name if they wrap systemd code, rather than only external
3261 code. Among them fsck@.service which is now
3262 systemd-fsck@.service.
3264 * The HaveWatchdog property has been removed from the D-Bus
3267 * systemd.confirm_spawn= on the kernel command line should now
3270 * There's a new man page crypttab(5) which details all options
3271 we actually understand.
3273 * systemd-nspawn gained a new --capability= switch to pass
3274 additional capabilities to the container.
3276 * timedated will now read known NTP implementation unit names
3277 from /usr/lib/systemd/ntp-units.d/*.list,
3278 systemd-timedated-ntp.target has been removed.
3280 * journalctl gained a new switch "-b" that lists log data of
3281 the current boot only.
3283 * The notify socket is in the abstract namespace again, in
3284 order to support daemons which chroot() at start-up.
3286 * There is a new Storage= configuration option for journald
3287 which allows configuration of where log data should go. This
3288 also provides a way to disable journal logging entirely, so
3289 that data collected is only forwarded to the console, the
3290 kernel log buffer or another syslog implementation.
3292 * Many bugfixes and optimizations
3294 Contributions from: Auke Kok, Colin Guthrie, Dave Reisner,
3295 David Strauss, Eelco Dolstra, Kay Sievers, Lennart Poettering,
3296 Lukas Nykryn, Michal Schmidt, Michal Sekletar, Paul Menzel,
3297 Shawn Landden, Tom Gundersen
3301 * "systemctl help <unit>" now shows the man page if one is
3304 * Several new man pages have been added.
3306 * MaxLevelStore=, MaxLevelSyslog=, MaxLevelKMsg=,
3307 MaxLevelConsole= can now be specified in
3308 journald.conf. These options allow reducing the amount of
3309 data stored on disk or forwarded by the log level.
3311 * TimerSlackNSec= can now be specified in system.conf for
3312 PID1. This allows system-wide power savings.
3314 Contributions from: Dave Reisner, Kay Sievers, Lauri Kasanen,
3315 Lennart Poettering, Malte Starostik, Marc-Antoine Perennou,
3320 * logind is now capable of (optionally) handling power and
3321 sleep keys as well as the lid switch.
3323 * journalctl now understands the syntax "journalctl
3324 /usr/bin/avahi-daemon" to get all log output of a specific
3327 * CapabilityBoundingSet= in system.conf now also influences
3328 the capability bound set of usermode helpers of the kernel.
3330 Contributions from: Daniel Drake, Daniel J. Walsh, Gert
3331 Michael Kulyk, Harald Hoyer, Jean Delvare, Kay Sievers,
3332 Lennart Poettering, Matthew Garrett, Matthias Clasen, Paul
3333 Menzel, Shawn Landden, Tero Roponen, Tom Gundersen
3337 * Note that we skipped 139 releases here in order to set the
3338 new version to something that is greater than both udev's
3339 and systemd's most recent version number.
3341 * udev: all udev sources are merged into the systemd source tree now.
3342 All future udev development will happen in the systemd tree. It
3343 is still fully supported to use the udev daemon and tools without
3344 systemd running, like in initramfs or other init systems. Building
3345 udev though, will require the *build* of the systemd tree, but
3346 udev can be properly *run* without systemd.
3348 * udev: /lib/udev/devices/ are not read anymore; systemd-tmpfiles
3349 should be used to create dead device nodes as workarounds for broken
3352 * udev: RUN+="socket:..." and udev_monitor_new_from_socket() is
3353 no longer supported. udev_monitor_new_from_netlink() needs to be
3354 used to subscribe to events.
3356 * udev: when udevd is started by systemd, processes which are left
3357 behind by forking them off of udev rules, are unconditionally cleaned
3358 up and killed now after the event handling has finished. Services or
3359 daemons must be started as systemd services. Services can be
3360 pulled-in by udev to get started, but they can no longer be directly
3361 forked by udev rules.
3363 * udev: the daemon binary is called systemd-udevd now and installed
3364 in /usr/lib/systemd/. Standalone builds or non-systemd systems need
3365 to adapt to that, create symlink, or rename the binary after building
3368 * libudev no longer provides these symbols:
3369 udev_monitor_from_socket()
3370 udev_queue_get_failed_list_entry()
3371 udev_get_{dev,sys,run}_path()
3372 The versions number was bumped and symbol versioning introduced.
3374 * systemd-loginctl and systemd-journalctl have been renamed
3375 to loginctl and journalctl to match systemctl.
3377 * The config files: /etc/systemd/systemd-logind.conf and
3378 /etc/systemd/systemd-journald.conf have been renamed to
3379 logind.conf and journald.conf. Package updates should rename
3380 the files to the new names on upgrade.
3382 * For almost all files the license is now LGPL2.1+, changed
3383 from the previous GPL2.0+. Exceptions are some minor stuff
3384 of udev (which will be changed to LGPL2.1 eventually, too),
3385 and the MIT licensed sd-daemon.[ch] library that is suitable
3386 to be used as drop-in files.
3388 * systemd and logind now handle system sleep states, in
3389 particular suspending and hibernating.
3391 * logind now implements a sleep/shutdown/idle inhibiting logic
3392 suitable for a variety of uses. Soonishly Lennart will blog
3393 about this in more detail.
3395 * var-run.mount and var-lock.mount are no longer provided
3396 (which prevously bind mounted these directories to their new
3397 places). Distributions which have not converted these
3398 directories to symlinks should consider stealing these files
3399 from git history and add them downstream.
3401 * We introduced the Documentation= field for units and added
3402 this to all our shipped units. This is useful to make it
3403 easier to explore the boot and the purpose of the various
3406 * All smaller setup units (such as
3407 systemd-vconsole-setup.service) now detect properly if they
3408 are run in a container and are skipped when
3409 appropriate. This guarantees an entirely noise-free boot in
3410 Linux container environments such as systemd-nspawn.
3412 * A framework for implementing offline system updates is now
3413 integrated, for details see:
3414 http://freedesktop.org/wiki/Software/systemd/SystemUpdates
3416 * A new service type Type=idle is available now which helps us
3417 avoiding ugly interleaving of getty output and boot status
3420 * There's now a system-wide CapabilityBoundingSet= option to
3421 globally reduce the set of capabilities for the
3422 system. This is useful to drop CAP_SYS_MKNOD, CAP_SYS_RAWIO,
3423 CAP_NET_RAW, CAP_SYS_MODULE, CAP_SYS_TIME, CAP_SYS_PTRACE or
3424 even CAP_NET_ADMIN system-wide for secure systems.
3426 * There are now system-wide DefaultLimitXXX= options to
3427 globally change the defaults of the various resource limits
3428 for all units started by PID 1.
3430 * Harald Hoyer's systemd test suite has been integrated into
3431 systemd which allows easy testing of systemd builds in qemu
3432 and nspawn. (This is really awesome! Ask us for details!)
3434 * The fstab parser is now implemented as generator, not inside
3437 * systemctl will now warn you if .mount units generated from
3438 /etc/fstab are out of date due to changes in fstab that
3439 have not been read by systemd yet.
3441 * systemd is now suitable for usage in initrds. Dracut has
3442 already been updated to make use of this. With this in place
3443 initrds get a slight bit faster but primarily are much
3444 easier to introspect and debug since "systemctl status" in
3445 the host system can be used to introspect initrd services,
3446 and the journal from the initrd is kept around too.
3448 * systemd-delta has been added, a tool to explore differences
3449 between user/admin configuration and vendor defaults.
3451 * PrivateTmp= now affects both /tmp and /var/tmp.
3453 * Boot time status messages are now much prettier and feature
3454 proper english language. Booting up systemd has never been
3457 * Read-ahead pack files now include the inode number of all
3458 files to pre-cache. When the inode changes the pre-caching
3459 is not attempted. This should be nicer to deal with updated
3460 packages which might result in changes of read-ahead
3463 * We now temporaritly lower the kernel's read_ahead_kb variable
3464 when collecting read-ahead data to ensure the kernel's
3465 built-in read-ahead does not add noise to our measurements
3466 of necessary blocks to pre-cache.
3468 * There's now RequiresMountsFor= to add automatic dependencies
3469 for all mounts necessary for a specific file system path.
3471 * MountAuto= and SwapAuto= have been removed from
3472 system.conf. Mounting file systems at boot has to take place
3475 * nspawn now learned a new switch --uuid= to set the machine
3476 ID on the command line.
3478 * nspawn now learned the -b switch to automatically search
3481 * vt102 is now the default TERM for serial TTYs, upgraded from
3484 * systemd-logind now works on VT-less systems.
3486 * The build tree has been reorganized. The individual
3487 components now have directories of their own.
3489 * A new condition type ConditionPathIsReadWrite= is now available.
3491 * nspawn learned the new -C switch to create cgroups for the
3492 container in other hierarchies.
3494 * We now have support for hardware watchdogs, configurable in
3497 * The scheduled shutdown logic now has a public API.
3499 * We now mount /tmp as tmpfs by default, but this can be
3500 masked and /etc/fstab can override it.
3502 * Since udisks does not make use of /media anymore we are not
3503 mounting a tmpfs on it anymore.
3505 * journalctl gained a new --local switch to only interleave
3506 locally generated journal files.
3508 * We can now load the IMA policy at boot automatically.
3510 * The GTK tools have been split off into a systemd-ui.
3512 Contributions from: Andreas Schwab, Auke Kok, Ayan George,
3513 Colin Guthrie, Daniel Mack, Dave Reisner, David Ward, Elan
3514 Ruusamäe, Frederic Crozat, Gergely Nagy, Guillermo Vidal,
3515 Hannes Reinecke, Harald Hoyer, Javier Jardón, Kay Sievers,
3516 Lennart Poettering, Lucas De Marchi, Léo Gillot-Lamure,
3517 Marc-Antoine Perennou, Martin Pitt, Matthew Monaco, Maxim
3518 A. Mikityanskiy, Michael Biebl, Michael Olbrich, Michal
3519 Schmidt, Nis Martensen, Patrick McCarty, Roberto Sassu, Shawn
3520 Landden, Sjoerd Simons, Sven Anders, Tollef Fog Heen, Tom
3525 * This is mostly a bugfix release
3527 * Support optional initialization of the machine ID from the
3528 KVM or container configured UUID.
3530 * Support immediate reboots with "systemctl reboot -ff"
3532 * Show /etc/os-release data in systemd-analyze output
3534 * Many bugfixes for the journal, including endianness fixes and
3535 ensuring that disk space enforcement works
3537 * sd-login.h is C++ comptaible again
3539 * Extend the /etc/os-release format on request of the Debian
3542 * We now refuse non-UTF8 strings used in various configuration
3543 and unit files. This is done to ensure we do not pass invalid
3544 data over D-Bus or expose it elsewhere.
3546 * Register Mimo USB Screens as suitable for automatic seat
3549 * Read SELinux client context from journal clients in a race
3552 * Reorder configuration file lookup order. /etc now always
3553 overrides /run in order to allow the administrator to always
3554 and unconditionally override vendor supplied or
3555 automatically generated data.
3557 * The various user visible bits of the journal now have man
3558 pages. We still lack man pages for the journal API calls
3561 * We now ship all man pages in HTML format again in the
3564 Contributions from: Dave Reisner, Dirk Eibach, Frederic
3565 Crozat, Harald Hoyer, Kay Sievers, Lennart Poettering, Marti
3566 Raudsepp, Michal Schmidt, Shawn Landden, Tero Roponen, Thierry
3571 * This is mostly a bugfix release
3573 * systems lacking /etc/os-release are no longer supported.
3575 * Various functionality updates to libsystemd-login.so
3577 * Track class of PAM logins to distuingish greeters from
3580 Contributions from: Kay Sievers, Lennart Poettering, Michael
3585 * This is an important bugfix release for v41.
3587 * Building man pages is now optional which should be useful
3588 for those building systemd from git but unwilling to install
3591 * Watchdog support for supervising services is now usable. In
3592 a future release support for hardware watchdogs
3593 (i.e. /dev/watchdog) will be added building on this.
3595 * Service start rate limiting is now configurable and can be
3596 turned off per service. When a start rate limit is hit a
3597 reboot can automatically be triggered.
3599 * New CanReboot(), CanPowerOff() bus calls in systemd-logind.
3601 Contributions from: Benjamin Franzke, Bill Nottingham,
3602 Frederic Crozat, Lennart Poettering, Michael Olbrich, Michal
3603 Schmidt, Michał Górny, Piotr Drąg
3607 * The systemd binary is installed /usr/lib/systemd/systemd now;
3608 An existing /sbin/init symlink needs to be adapted with the
3611 * The code that loads kernel modules has been ported to invoke
3612 libkmod directly, instead of modprobe. This means we do not
3613 support systems with module-init-tools anymore.
3615 * Watchdog support is now already useful, but still not
3618 * A new kernel command line option systemd.setenv= is
3619 understood to set system wide environment variables
3620 dynamically at boot.
3622 * We now limit the set of capabilities of systemd-journald.
3624 * We now set SIGPIPE to ignore by default, since it only is
3625 useful in shell pipelines, and has little use in general
3626 code. This can be disabled with IgnoreSIPIPE=no in unit
3629 Contributions from: Benjamin Franzke, Kay Sievers, Lennart
3630 Poettering, Michael Olbrich, Michal Schmidt, Tom Gundersen,
3635 * This is mostly a bugfix release
3637 * We now expose the reason why a service failed in the
3638 "Result" D-Bus property.
3640 * Rudimentary service watchdog support (will be completed over
3641 the next few releases.)
3643 * When systemd forks off in order execute some service we will
3644 now immediately changes its argv[0] to reflect which process
3645 it will execute. This is useful to minimize the time window
3646 with a generic argv[0], which makes bootcharts more useful
3648 Contributions from: Alvaro Soliverez, Chris Paulson-Ellis, Kay
3649 Sievers, Lennart Poettering, Michael Olbrich, Michal Schmidt,
3650 Mike Kazantsev, Ray Strode
3654 * This is mostly a test release, but incorporates many
3657 * New systemd-cgtop tool to show control groups by their
3660 * Linking against libacl for ACLs is optional again. If
3661 disabled, support tracking device access for active logins
3662 goes becomes unavailable, and so does access to the user
3663 journals by the respective users.
3665 * If a group "adm" exists, journal files are automatically
3666 owned by them, thus allow members of this group full access
3667 to the system journal as well as all user journals.
3669 * The journal now stores the SELinux context of the logging
3670 client for all entries.
3672 * Add C++ inclusion guards to all public headers
3674 * New output mode "cat" in the journal to print only text
3675 messages, without any meta data like date or time.
3677 * Include tiny X server wrapper as a temporary stop-gap to
3678 teach XOrg udev display enumeration. This is used by display
3679 managers such as gdm, and will go away as soon as XOrg
3680 learned native udev hotplugging for display devices.
3682 * Add new systemd-cat tool for executing arbitrary programs
3683 with STDERR/STDOUT connected to the journal. Can also act as
3684 BSD logger replacement, and does so by default.
3686 * Optionally store all locally generated coredumps in the
3687 journal along with meta data.
3689 * systemd-tmpfiles learnt four new commands: n, L, c, b, for
3690 writing short strings to files (for usage for /sys), and for
3691 creating symlinks, character and block device nodes.
3693 * New unit file option ControlGroupPersistent= to make cgroups
3694 persistent, following the mechanisms outlined in
3695 http://www.freedesktop.org/wiki/Software/systemd/PaxControlGroups
3697 * Support multiple local RTCs in a sane way
3699 * No longer monopolize IO when replaying readahead data on
3700 rotating disks, since we might starve non-file-system IO to
3701 death, since fanotify() will not see accesses done by blkid,
3704 * Do not show kernel threads in systemd-cgls anymore, unless
3705 requested with new -k switch.
3707 Contributions from: Dan Horák, Kay Sievers, Lennart
3708 Poettering, Michal Schmidt
3712 * This is mostly a test release, but incorporates many
3715 * The git repository moved to:
3716 git://anongit.freedesktop.org/systemd/systemd
3717 ssh://git.freedesktop.org/git/systemd/systemd
3719 * First release with the journal
3720 http://0pointer.de/blog/projects/the-journal.html
3722 * The journal replaces both systemd-kmsg-syslogd and
3723 systemd-stdout-bridge.
3725 * New sd_pid_get_unit() API call in libsystemd-logind
3727 * Many systemadm clean-ups
3729 * Introduce remote-fs-pre.target which is ordered before all
3730 remote mounts and may be used to start services before all
3733 * Added Mageia support
3735 * Add bash completion for systemd-loginctl
3737 * Actively monitor PID file creation for daemons which exit in
3738 the parent process before having finished writing the PID
3739 file in the daemon process. Daemons which do this need to be
3740 fixed (i.e. PID file creation must have finished before the
3741 parent exits), but we now react a bit more gracefully to them.
3743 * Add colourful boot output, mimicking the well-known output
3744 of existing distributions.
3746 * New option PassCredentials= for socket units, for
3747 compatibility with a recent kernel ABI breakage.
3749 * /etc/rc.local is now hooked in via a generator binary, and
3750 thus will no longer act as synchronization point during
3753 * systemctl list-unit-files now supports --root=.
3755 * systemd-tmpfiles now understands two new commands: z, Z for
3756 relabelling files according to the SELinux database. This is
3757 useful to apply SELinux labels to specific files in /sys,
3760 * Output of SysV services is now forwarded to both the console
3761 and the journal by default, not only just the console.
3763 * New man pages for all APIs from libsystemd-login.
3765 * The build tree got reorganized and a the build system is a
3766 lot more modular allowing embedded setups to specifically
3767 select the components of systemd they are interested in.
3769 * Support for Linux systems lacking the kernel VT subsystem is
3772 * configure's --with-rootdir= got renamed to
3773 --with-rootprefix= to follow the naming used by udev and
3776 * Unless specified otherwise we will now install to /usr instead
3777 of /usr/local by default.
3779 * Processes with '@' in argv[0][0] are now excluded from the
3780 final shut-down killing spree, following the logic explained
3782 http://www.freedesktop.org/wiki/Software/systemd/RootStorageDaemons
3784 * All processes remaining in a service cgroup when we enter
3785 the START or START_PRE states are now killed with
3786 SIGKILL. That means it is no longer possible to spawn
3787 background processes from ExecStart= lines (which was never
3788 supported anyway, and bad style).
3790 * New PropagateReloadTo=/PropagateReloadFrom= options to bind
3791 reloading of units together.
3793 Contributions from: Bill Nottingham, Daniel J. Walsh, Dave
3794 Reisner, Dexter Morgan, Gregs Gregs, Jonathan Nieder, Kay
3795 Sievers, Lennart Poettering, Michael Biebl, Michal Schmidt,
3796 Michał Górny, Ran Benita, Thomas Jarosch, Tim Waugh, Tollef
3797 Fog Heen, Tom Gundersen, Zbigniew Jędrzejewski-Szmek